Cogent Communications Holdings Inc Aktie

Always pleased to have Dave Schaeffer from Cogent here with us. He's kind enough to come every year and great to hear the latest from you, Dave. So thank you again.

Michael, thank you for hosting me. As always, thank BofA for a great venue and thank investors for taking time out of their day to hear what we want to talk about.

Outstanding. So I had a number of questions that I want to go through. I will leave a few minutes at the end if there are any questions from the audience. So when I mentioned that, there should be microphones going around, just raise your hand and we'll take time for that. So Dave, I think you would admit that growth has been disappointing and below your own projections for the last year or so due to a number of factors. So -- can you just remind us maybe what went right and what went wrong or maybe not wrong, but didn't meet expectations in the last 12 or 18 months? And then why you believe that's going to change in the next 12 or 18 months?

So let me start by laying kind of the backdrop. Cogent had done no acquisitions between June of 2005 and May of 2023. In that 18-year period, we had a compounded average growth rate of 10.2% a year. Our corporate segment actually was outperforming that growing at 11.2% a year over that 18-year period. And because the majority of our sales, 76% of our revenues were on-net services, we were experiencing consistent margin expansion. Our EBITDA margins expanded at approximately 220 basis points a year.

So the on-net, it's about 90% incremental margin to growth, right? Correct.

No, the actual total margins had expanded. So we went public in June of '05 with 0 EBITDA margins. When we acquired Sprint, our EBITDA margins were 40.5%. We were acquiring a business that was in decline and burning cash. The business we acquired for the 3 previous years had declined at 10.9% a year, and it represented 42% of the revenues of the combined company. So we expected our total revenue growth to go negative. We also knew that the company we acquired was operating a number of unprofitable products and network services and had a negative 60% EBITDA margin. We outlined a plan to take costs out -- and more importantly, we had a program to repurpose the physical assets of that business that have been sitting dormant since 2015 and turn that old TDM voice network that was built at a capital cost of over $20 billion in 1980s into a modern wavelength network to serve international carriers, regional carriers and content distributors.

We did not anticipate the tailwind of AI. So that was a bluebird. I will not profess. I was Carvoyant in September of '22 when we signed the deal, we knew that AI was going to burst upon the scenes and create a $7 trillion capital investment cycle. So what went right? The ability to take costs out were actually better than we expected. We actually raised our cost reduction targets, and we achieved those cost reductions quicker than we expected. What went wrong was we lost more of that revenue than we expected. Now in a way, that's a blessing in the skies because that revenue was negative margin. Today, the Sprint customer base is margin 0. So it doesn't help, but it doesn't hurt. It has shrunk from 42% of our total revenues to only 15% of our revenues. We today have a business that is at least breakeven in those Sprint customers.

The second part of the transaction was the establishment of a new business, a wavelength business. We actually reenabled the Sprint network on budget and on time. We did not exceed our goals, but we did not fall short. We also made the election after the transaction was announced to invest an additional $100 million into the physical buildings that we acquired and convert those former switch sites to data centers. We had a program to divest of 24 of the 125 that we did convert. We chose those based on their location, size and proximity to major markets. We then completed that.

There was a great deal of skepticism that the company already was spending money on building a wave network and then committing an additional $100 million into what was viewed as a speculative endeavor. We, I think, proved that investment to be correct. We sold 10 of those 125 facilities for $225 million to $24.1 million net because there was very little transaction costs associated with that divestiture. We did that in-house and we used those proceeds to reduce debt on our balance sheet. So I think on the data center monetization, it was not originally anticipated. We did spend more than we expected, but we got more than we spent back. And we still have 14 data centers that we plan to divest of using those proceeds to further delever and we will be left with the remaining footprint that while we view it as not monetizable through sale, we should be able to lease out that capacity.

And Dave, just quickly, on the 14 remaining data centers, do they look like the 10 that you already sold, meaning location, occupancy? Should we expect similar valuation per megawatt for the remaining 14?

We have never sold a data center prior to this. We had no real view on how well we will do other than we felt that we would do much better than the $100 million invested. We ended up getting $4.3 million a megawatt. The 10 facilities we sold were about 53 megawatts getting to the 225. The remaining facilities are 54 megawatts, so about equivalent. And at least from our perspective, work relatively similar. I think the expectations should be about the same. Each of the facilities has some different characteristics, whether it be physical site or availability of incremental power almost all of these facilities are empty with no real revenue associated with them. We have the added benefit that when we divested of the first 10, we reduced a negative drag on our EBITDA by $7 million a year.

We were effectively paying the real estate taxes, personal property taxes and maintenance of facilities with no revenue.

And these are old wire centers, right? They were not purpose-built data centers, old wire centers, you converted with $100 million to make them into data centers. So they're different maybe than a traditional data center.

That's correct.

Correct.

What these were if you look at a typical facility, it was 15 miles from a downtown was 6 or 7 acres with a 40,000-foot building with 4 to 5 megawatts of inbound power full of old Class 4 and Class 5 telephone switches. There were 23,800 cabinets of telephone switch gear that we removed from the footprint and disposed of. We had to convert these facilities from negative 48-volt DC power plant, which is how a switch site operates to AC120, which is how a data center operates. We had to refurbish the air conditioning, the fire suppression systems, the security systems and do some cosmetic refurbishment to then make these basic data centers. These are still facilities that have only about 100 watts per square foot and are not today able to support high-density liquid-cooled cabinets.

Now they are modifiable to that, but they are smaller, much more distributed than the types of data centers that are in the headlines today that people are protesting where people are trying to stop the construction of a 1 gigawatt million square foot facility on 1,000 acres of land. These are much more innocuous. And they, I think, will have a great deal of applicability for inference. In fact, I square, the purchaser of the first 10, did not disclose to us during the purchase process, what their ultimate plan was. But subsequent to buying them and now standing up a management team to start to commercialize these facilities, they disclosed to us that their ultimate plan is to make these inference facilities and kind of have token repositories distributed around the country where they can then be providing inference as a service to Neo clouds and hyperscalers.

They ultimately tend to invest $1 billion at least that's what their public statement is. $225 million of that came to us. Now to what else went right and wrong. I want to finish the answer to your question. Cogent's core business actually accelerated. 84% of our revenues come from selling Internet-based services. That business had been negatively impacted by the pandemic. It had seen its growth rate fall from that 10.2% down to 5% pre-acquisition. In the 3 years since the acquisition, the core Cogent business has actually seen its growth rate reaccelerate to about 9%. That business over a 3-year period grew 29%.

The acquired Sprint business rate of decline accelerated and declined in totality at 69%. So there is both good and bad post transaction in our 2 customer bases. Probably the biggest area of disappointment has been in our wavelength business, our brand new business. We knew that it was going to take close to 1.5 years to 2 years to do the conversion. We actually started on that conversion prior to the acquisition after the deal was announced. And we completed the enablement of an initial footprint of 800 data centers by December 31, 2024. So about 18 months after the deal was closed, right on plan, right on budget. Where we were disappointed as we hoped to be able to sell more wavelength services on an interim basis as we were bringing these facilities online. We sold about 1,000 waves in that interim period, not the 3,000 that we had hoped to.

Then once the entire footprint was enabled, we have continued to grow that business. Last quarter, it grew 62% year-over-year. It grew 9.2% on a sequential basis and has performed well, but off of a small base. We today have 3% market share in the North American intercity wavelength market. Our ultimate goal is to get to 25% share. We were hoping to do that by midyear '28. I don't believe sitting here in the fall of '26, that is realistic. I think the ultimate goal is realistic, but the pace to get there is going to take longer. And I think the 2 reasons why our ramp in revenue growth was slower than we anticipated were, one, our need to build credibility in this new product. I think we had earned and have a great deal of credibility in our core product suite, but I think many customers did not associate Cogent with wavelengths and were comprehensive about our ability to deliver the quality that we represented.

I think we've solved that problem. We have now sold wavelengths to 548 unique customers. We've delivered service in 608 of the ultimate 1,137 locations that we brought online. I think the second challenge has been the supply chain challenges in the entire technology and telecom ecosystem that we did not anticipate. Now on the positive side, we get the tailwind of a new incremental use case for wavelengths that is AI training, connecting a data source to a training facility and doing it with defined latency. While that is more expensive and harder to administer than using the Internet, it does allow for more optimal utilization of the GPU capacity. Since 60% of the capital cost of a training facility is in the GPUs, maximizing their efficiency is critical. And with that...

How does it maximize the efficiency of the GPUs?

So, every large language model works on the same basic architecture. They each have variances. They each do it slightly different, and they are evolving. But what they do is take a large set of data that was collected over the Internet, literally exabytes of data. They tokenize that data, breaking words into recognizable tokens that GPUs can process. They then pose a question to that set of data that they already know the answer to. And the GPUs crank through that data multiple times and develop a pattern upon which the predicted answer and the actual answer are matched. They do that recursively, repeatedly.

And over time, they build a neural network that can then be applied to a new set of data. Most of the data storage exists in preexisting data centers and major data collection points, Northern Virginia, Ashburn, San Jose, Greater Chicago market, Greater Dallas market, Miami market, Amsterdam in the Netherlands, the U.K., mostly in England, Frankfurt. And as a result, there's a couple of dozen locations globally where most data sits. Most of the training is occurring in locations that are remote from those locations. There are locations where there is power and space available.

We're seeing new data centers being constructed in some very unobvious locations, Southern Louisiana, Central North Dakota, Western Wyoming, Western Texas. And they're going to those locations because power, land, water are available. So in order to train, you have to take the data out of the data center where it's stored, move it to where the GPU is and then return it back after the training is complete. And that process is best done over a wavelength because a wavelength is a deterministic point-to-point service. On a per bit mile basis, it's actually about 2.5x more expensive than using the public -- but if you use the public Internet to transmit that information, you would have to buffer that information or you would have to be able to degrade the productivity of your GPUs.

So all of the AI training is today done using wavelengths as an incremental demand set to what we expected from regional networks, from international carriers and content deliverers. And this trend is going to continue for multiple years, creating a huge amount of demand. Now what that training demand has done, though, is tax the entire supply chain, whether it be power generation, whether it be power transmission, whether it be land, whether it be cooling systems, switchgear, fire protection. At every level, this surge in demand has created shortages. There's a tremendous memory shortage where prices have gone up by a factor of 5. Telecom equipment, which would typically be orderable and deliverable in 90 days, has now got a 3-year delivery window.

You've got a new class of buyers at a traditional telecom equipment supplier, 90% of their revenues were coming from service providers like Cogent or AT&T or Deutsche Telekom. Now 60% of their demand comes from 1 of 4 major hyperscalers. This is a shock to the system that has created a number of supply chain constraints. That impacts us 2 ways. One, it's raised prices for equipment that Cogent and other service providers buy. It's created telecom equipment inflation, which is unheard of for 50 years, price performance had consistently improved. This is the first time in the digital age where price per unit of productivity is going up. That will get reverted and we will return to the Moore's Law type of price declines, but that's probably going to take a couple of years to work through all of the systematic constraints.

Can I ask you, Dave, before I forget. You mentioned 3% share today, hoping to get to around 25% over time. And obviously, AI has been unexpected, right, positive incremental. How much is AI training spending on wavelength today in general -- in general, how much are they spending today?

Today, AI training, which is coming both from hyperscalers and Neo clouds, represents about 1/4 of the total market, but it's the fastest-growing segment. It is not always easily measurable because the major hyperscalers also operate traditional cloud distribution businesses, and they typically buy wavelengths through a centralized purchasing mechanism and don't always disclose to Cogent or any other supplier, what the ultimate use of that wave is. So when a Microsoft or a Google or a Meta or an Amazon buys a wavelength, they don't put a little check mark and see.

Yes, you don't know exactly how they're using it. You can't differentiate. Is your market share of that 25%, is that in line with your 3% share? Or is it higher?

I think it's a little higher with that group of customers for 2 reasons. One, for the existing hyperscalers, we had long-standing 15-plus year relationships and have been able to leverage those. Secondly, for the Neo clouds, our ability to have such a broad footprint and rapid deployment schedule is differentiated.

[indiscernible] Seeing is much faster than others, I think, for wavelength, correct? Yes.

Yes. So that really helps us with someone like an Anthropic or an OpenAI or a CoreWeave or Nebius or Gore or Lambda Labs, this next wave of companies that each have slightly different business models but are really only in existence for AI training at this point.

And so let's just say that just round numbers, say your 10% share of training wavelength traffic or spending, who ask you other 90%? And do they have a 90% because they're building connectivity, providing wavelengths to these out-of-the way AI training facilities that you never had or will connect to because they're not part of that number of data centers that you're providing wavelengths. Is that why? And then who has that 90% that you don't have, if I have that number roughly right?

So that is not correct. So we today go to 59 single-tenant facilities. Cogent is willing to build to a facility but needs to have sufficient commitments from that single tenant. There are 4 ways in which we can get to a single-tenant facility. The first is we just build there. We're happy to do that if either the customer signs a long enough and large enough multiyear contract or they sign a shorter contract and give us enough of a nonrecurring payment to cover our capital expenditure because if we build to a single-tenant facility and for whatever reason, they stop using us, whether their business model fails or the facility becomes dormant or they switch providers, we then have a completely stranded asset.

The second way we can do that is, in many cases, these companies have already built their own fiber as part of developing these facilities back to a major aggregation point or point where they can splice into our network. Our network is probably one of the most diverse in terms of physical routes and has more splice points available to it. So we are in many cases, connecting to those network locations by the customers' own fiber. The third model is we buy fiber. So Cogent's core business had always been predicated on buying IRUs, indefeasible rights of use. We have purchased 94,000 route miles of intercity fiber 34,000 route miles of metropolitan fiber from 383 different suppliers around the world.

We then take that dark fiber and put our own optronics on it. Oftentimes, when we go to a remote location, we may piece together 2, 3 different fiber purchases to get very close to that location and then do literally maybe the last 100 yards, last half a mile into the facility. So we also have that ability to use our network of providers to allow us to get close to these facilities. And then the final option is for the data center owner to buy with services from someone else back to a carrier neutral to then ride our network. That is more complex, but it's also more desirable. So the Internet, by its very design is resilient. It constantly reroutes traffic around problems. That's the fundamental design of the Internet. A wavelength is an unprotected point-to-point service, which means any physical interruption along that path and the service fails to work.

Almost no customer buys wavelengths as a single service. They almost always buy a primary and a backup wavelength. In the ideal case, they would like to have the same 2 city pairs, identical latency, but complete vendor and complete path diversity. That ideal really exists in the real world. So what you do is get as much diversity as you can. So an example would be if I'm going from Western Nebraska back in the Denver, I may be able to get diversity 80% of the way, but there may be an Indian reservation I connect through in which there's only one physical fiber pass. So I have a pinch point where part of the math will have a common point of failure. Then when you connect in Denver, you would like to ride a network that's completely diverse from 15th Street, that's a major aggregation point in Denver to, say, 707 Wilshire, a major aggregation point in L.A.

There, you can get a completely independent wave, and you might buy one wave from one vendor and a second wave from us. So I think Cogent actually has done quite well with customers that buy waves from others and using us as diversity. And conversely, people that buy from us as primary, we use other vendors for that diversity. The final point that makes this even a little more complex is that customers sometimes substitute dark fiber for wavelengths as a fungible product. It is yet again more expensive. It is harder to implement. It has a longer lead time, but it may be the only way to serve some of these remote locations because no one would build to that location speculatively. And in those cases, they may like that dark fiber and then look to buy a wavelength to protect as much of that physical path as they can.

So the wavelength market continues to grow and the historic use cases are pretty static. It's really AI that's driving the growth.

Okay. I have 1 or 2 more, but I do want to open up to the audience if there are any questions. I know we have a microphone that can go around. Just raise your hand if you have any questions. And while we're waiting for that, I wanted to go back to the question. I really -- really explain that to me. So who has the most market share than in wavelengths, if my 10% estimate was even roughly right. Who makes up that 90%? And why are they being selected? And then second part to the question is how have AT&T and Verizon now emphasizing this market as well as a growth area? How has that impacted your business and your wins?

So in the intercity market, Lumen is the dominant player. Zayo is the second largest player. Those 2 companies became dominant players by rolling up many other network players gaining market share. It has been a fairly concentrated market, and it's a market that historically, AT&T and Verizon did not participate in. They have recently announced that they would like to get back into that market, but have not yet made meaningful investments or inroads into that market.

Verizon has had an announcement or 2 though, right, with some partnerships with hyperscalers.

So they are -- Verizon at least has announced a new build for a hyperscaler along with the sale of some of their inventory.

Correct.

Verizon is actually one of our 383 counterparties we buy select fiber from. So they're willing to sell the fiber, whether it's to us as a service provider or a hyperscaler but they are also willing to do brand-new builds. So to remind investors, about a decade ago, Verizon announced a program called One Fiber, and they built out metropolitan fiber networks in 69 markets out of footprint, which was really unprecedented for a phone company. They spent about $3.5 billion building that network, primarily to support their wireless backhaul requirements. They have subsequently tried to monetize that by selling wavelengths. So Verizon is a wavelength seller in the metro nationally. AT&T is a dominant seller in its footprint, but does not sell wavelengths in the metro out of footprint.

Cogent is not focused on that metro market because we have no real competitive advantage. Our primary competitive advantage is in the intercity market. So if we look at the total North American market at about $3.5 billion, roughly 150,000 discrete wavelengths. The dominant players in the metro are Verizon and AT&T second, followed by Zayo third, Lumen in a distant fourth and Cogent way, way behind. . And then on the intercity piece, AT&T and Verizon, actually Verizon's predecessor MCI, initially developed an intercity transport network using microwave they converted to fiber in the late '90s, but they were burned twice, first by switch-based resellers who all went bankrupt in the LED market and then by regional ISPs.

At that point, both AT&T and MCI exited the wavelength business and really left the market to the new entrant. So it was Broadwing, it was Williams, it was Quest, it was Level 3. And over time, Lumen has consolidated many of them. Zayo's consolidation started in a different path. They bought a bunch of local metro markets and eventually link them together. 2 of them have been the dominant suppliers of intercity wavelengths and where Cogent has an advantage as we come to market with routes that those carriers don't have and therefore, provide diversity that they cannot provide.

Okay. That's very helpful. We have about 1.5 minutes left. So I'm going to leave it for one more question, Dave. I want to keep wavelengths here for a second. You kind of an part earlier, but part of what you attribute the slower growth to earlier in the year was you said Cogent's provisioning service so fast and customers just can't take it or accept it. Are you still seeing that issue where customers not accepting delivery once you provisioned it? Or is that time shrinking?

We are still seeing that issue, and I think it will continue for some time.

That's the equipment issue or -- because there was some question about a customer we're even just maybe overprovisioning or overbuying wavelengths, not knowing exactly who is going to deliver and who wasn't. And then because you weren't forcing to take delivery, there was no penalty, right? So there was no risk in overbuying if that is, in fact, what they were doing.

So there is actually a penalty. These contracts are take-or-pay contracts. But as a new entrant into the market, we are careful about when we force bill customers and pressure them into taking something that they would like to delay. Customers are not always transparent with us on the reasons for the delay. And in many cases, the customers are at the mercy of their suppliers. They signed a lease to be in a data center that was supposed to open in June, and it's not going to open until November or they go to a data center and the data center operator says, I'm capped on power now and won't have incremental power available for sale for 4 more months. or they may find that they're behind in receiving their GPUs, they're behind in receiving servers. And it could be a combination of multiple constraints.

So all of these things have caused customers to do 2 things: to delay and also be more agile than normal. We've had dozens of cases where customers bought a service to a data center in a market. We provisioned it, and they came back to us and said, hold off, we need you to move to another data center in the same market because the first data center was out of capacity for us. Now I do think that as the supply chain issues get resolved and as the industry matures, these problems will dissipate. We're in a situation where $7 trillion of capital has been announced and only $1 trillion of it has been spent.

That was excellent, Dave. We're out of time. Thank you so much. Once again, I appreciate it.

Thanks, Michael.Thank you all very much.

Thank you again, Dave.

For those of you I haven't met, I'm Mike Rollins. I cover communication services and infrastructure for Citi. And just before we begin, disclosures are available at the registration desk. It's a real pleasure to welcome back to the conference, Dave Schaeffer, CEO of Cogent. Dave, thank you so much for being with us. It's great to see you.

Mike, thanks for hosting me. Thank you, Roberta. Thanks Citi for a great venue. Thank investors for hanging around late in the day. And I'm here to try to answer some questions.

Great. I also want to introduce my colleague, Roberta Versiani, and we're going to co-host today's session. So maybe to kick us off, so Dave, when you look at the outlook over time, look at the growth trajectory. How much comes from the more recent product expansions? So, Waves, I mean put IPv4 into that bucket versus your traditional transit and VPN businesses, whether it's NetCentric or corporate?

Yes. So 84% of Cogent's revenues come from our legacy product set. That group of products is continuing to grow and has grown organically since the company began selling services in 2002. So for 24 years, we've experienced organic growth. Our DIA business is focused on large multi-tenant office buildings. That business had grown historically for an 18-year period at about 11.2% per year. When the pandemic hit, the growth rate in that corporate segment decelerated to a negative growth rate of 9%. It has subsequently rebounded to about a 3% growth rate.

We today have about 35% market share in the 1.1 billion square feet that we have directly on-net where we are an ISP. The 2 primary products we sell are Internet access and VPN services, mostly based on a VPLS platform, some based on legacy MPLS platforms. We also have a transit business. We operate in 310 markets around the world, 58 countries and 1,950 data centers. I was actually asked the question earlier today, what is the power footprint of that universe of data centers? It's about 17.5 gigawatts of power in that footprint.

Our primary product in that market is selling metered Internet access. It accounts for 98% of the traffic on Cogent's IP network. It is a business that today in revenue is growing about 8%. It is a business that traditionally is driven by a specific application at a point in time.

So even though Internet traffic has grown at a compounded rate of 23% per year over the Internet's 30-year history, it's been very lumpy, and you can look at 6 discrete waves of growth. We're actually experiencing acceleration in that traffic growth now coming from agentic AI. In the past 2 quarters, we have seen traffic growth accelerate from 8% year-over-year to 15% year-over-year, allowing us to drive that high single-digit revenue growth.

Now for the newer parts of Cogent's business. We have leased out IP address space since 2015. Between 2015 and 2022, we only leased addresses to companies that bought bandwidth from us. And it was a relatively small business. In '22, that business did about $20 million in revenue. Because of the headwinds we were facing from the pandemic, we lifted the restriction on selling bandwidth only or selling addresses only to customers that bought bandwidth from us. And we saw a rapid acceleration in that business. Our IPv4 leasing business today is about 7% of total revenues. It is roughly a $70 million business. We have leased out about 40% of the inventory of addresses that we have. And it's a business that last year on a year-over-year basis, grew 18%.

We recently entered the wavelength business. Prior to acquiring Sprint, Sprint was not in the wavelength business. Cogent was not in the wavelength business. We saw this as an adjacent market where our technical acumen, coupled with our sales and customer relationships would allow us to address a new $2 billion total addressable market for North America. We spent nearly 2 years reconfiguring the legacy Sprint Global Markets network that was built to carry TDM voice into a wavelength network.

In earnest, we began selling those services in January of '25. With that sale, we have seen our wavelength business grow. And today, it is about a $65 million run-rate business, about 6.5% of Cogent's total revenues. It grew on a year-over-year basis, 62%. And in the last quarter, it grew 9.2% sequentially.

To round out our product portfolio, we have a colocation business. It has always been a small part of Cogent's revenues. It represents about 2% of our revenue base. We today have 170 data centers, representing 1.3 million square feet and roughly 150 megawatts of power. We have 14 of those data centers up for sale. Last quarter, we completed the sale of 10 of these surplus data centers. That sale resulted in $225 million of cash proceeds. The actual transaction costs were less than $1 million. We were able to effectively shelter that gain from taxes and used $175 million of those proceeds to reduce our leverage.

Great. One more on just like the top line trends before we get more into the products in detail. So what's your latest thoughts on when Cogent can return to sequential revenue growth and then year-over-year revenue growth?

So, for 18 years prior to acquiring Sprint, but after going public in June of 2005, Cogent's total business grew at a compounded average rate of 10.2% a year with no acquisitions. We acquired the Sprint business, which represented 42% of the revenues of the combined company. Cogent represented 58%. Sprint's rate of revenue decline was 10.9% a year for the 3 previous years prior to deal closing. We actually took that 10.9% rate of decline and accelerated it to over 20%.

So in the 12 quarters post transaction, the combined company has had an average growth rate of a negative 5% top line growth. The Sprint revenues have declined from 42% of combined company's revenues to 15% of revenues and the aggregate rate of revenue decline in that acquired base in 3 years has been 69%. During the same period, the organic Cogent business has grown from 58% of revenues to 85% of revenues and had a compounded growth rate over that 3-year period of 29%. We were encouraged last fall when we actually had 2 out of 3 months in a quarter of positive growth. But we then had some more churn coming from the acquired Sprint customer base. We have been able to inflect the margin on that acquired base from a negative 60% EBITDA margin to 0.

So when we lose $1 of that acquired revenue, it no longer has any impact on our margin. But as we sell organic Cogent services, they carry much higher contribution margins. In the most recent quarter, 82% of our sales were on-net, 18% of sales were off-net. For the blended base, which still does include that acquired Sprint off-net base, we were 62% of on-net revenue, 38% off. As a result of this, the rate of revenue decline of the entire company is declining. It is difficult for me, Mike, to be able to give you a specific target date, but it does feel like that within the next year or 2, for sure, the Sprint business will be so de minimis and the organic Cogent businesses will be large enough that we will be able to return to total top line growth.

Even with that 5% negative top line growth, we have been able to grow EBITDA, not just in margin, but in absolute terms in 11 of the past 12 quarters. And the rate of margin expansion over the past 3 years has averaged almost 800 basis points a year, far better than Cogent's historical average. We expect that rate of margin expansion to begin to moderate and revert back to the roughly 200 basis points a year that Cogent has historically delivered.

Great. And on asset monetization, what's the status of your process to monetize your remaining data center assets? And are there any other assets that you might be looking to monetize in the future?

Yes, Roberta. So first of all, we inherited 482 buildings that comprise 1.9 million square feet and 230 megawatts of inbound power. None of these were data centers. Cogent had 54 data centers; 52 of those 54 were in leaseholds, and they had a total of 634,000 feet of raised floor and 69 megawatts of power. We initially did not plan to convert the Sprint switch sites to data centers. But by January of '24, it became obvious to us that the asset power that we had in place had value to the market.

We then initiated a program in June of '24 that ran 1 year, spent $100 million and converted 125 of those facilities into data centers. We identified 24 of the 125 as being large enough and geographically located in desirable locations, we should sell them. We placed those assets in the market. And on June 29 of this year, we closed the sale for the first 10 at $4.3 million a megawatt, resulting in the $225 million in proceeds; $125 million was a direct gain.

The remaining 14 are still in the market for sale. We have 2 LOIs that we have received from credible counterparties. We have not yet chosen to move forward. I do think we will be selling most, if not all, of those 14. Above and beyond that, we have other data centers that may be for sale, many of them smaller and more geographically dispersed, but could be interesting for edge applications. A second asset that we could elect to monetize is our excess IP address space. As I mentioned, we have leased out 40% of our inventory. We have 38 million addresses. We have approximately $15 million of them under lease. We took that leased revenue and securitized it with 2 asset-backed securitizations at a cost of capital of about 6.8%. We may elect to either sell or securitize revenues from the remaining 23 million addresses.

Finally, we have an intercity and metro dark fiber network. We are typically using 3 pairs or 6 fibers on those networks. We have strand counts from 24 to 144 fibers, and we may elect to sell some dark fiber. We have already done a handful of dark fiber sales. But to date, our primary focus has been on the data center monetization.

Dave, maybe just focusing on corporate for a moment. Historically, you and I have had a conversation about the marketing strategy, and you've outlined the importance of keeping it simple, keeping it low cost of acquisition and just driving penetration in your buildings with your Internet product and your VPNs. As that business sounds like that from our conversations, the penetration is maturing. Do you go into this next phase of trying to then sell more services into these customers? And are there services that you could sell to try to expand the TAM for the customers that you have versus the motion of just trying to deepen share in these corporate buildings?

So the answer is no. And I know that is against conventional wisdom. Part of what got Sprint in trouble was as its corporate and enterprise business declined, their solution was to sell more and more products to those customers further and further away from their core competency and eroding margins.

T-Mobile realized that, that was a losing strategy and evaluated either shutting down the business or selling it. While they paid Cogent $700 million, their all-in cost to divest of that asset was about $1.3 billion. And I would attribute most, if not all of that as a result of product diversification.

Now we do have the opportunity to upsell customers to larger connections. When you started covering Cogent, I guess Roberta was in elementary school at that time, and we were selling 100-meg connections that were viewed as far bigger than a customer could ever use with the average corporate customer using only about 1.2 megabits or less than 1.5% of the connection.

In the late, kind of, '18 or '19 timeframe, we began a program to upgrade those customers. And virtually all of our customers now take a 1-gig connection, 10x what they originally had, and that gave us the ability to uplift ARPUs. So you have 2 different things going on simultaneously. Price per unit falling, but bundle size is growing, pulling ARPUs up, but at a more moderate rate. We have begun another round of capacity augmentations.

We have enabled all 1,865 of our multi-tenant office buildings to also sell 10-gig connections. While today, only about 4% of our corporate base uses those 10-gig connections, they generally result in about 2.5x as much revenue per connection. Now the utilization rates fall even lower, but there is a subset of the market that just wants the best connection that money can buy. That is a far better strategy than us becoming a security provider, a hosting company, a managed service provider, a network as-a-service service provider, which I still struggle to fully understand what that means.

Great. And let's talk a little bit about Waves again. Are there any catalysts in the next 12 months that could accelerate the pace of Waves in volume and revenue dollars?

I believe so, Roberta. So we spent 2 years building a wave-optimized network out of a former TDM voice network. To do that, we had to reconfigure our metropolitan networks in over 110 cities. We deployed over 200 Wave rings in those markets and deployed about 110 reconfigurable add-drop multiplexers. We had initially hoped that we could sell Waves in only the largest data centers. We quickly found out that the Wave demand was far more dispersed.

We saw one of our competitors, a major household name phone company announce a program to sell intercity Waves in 8 data centers. And we haven't heard anything else about that after the press release came out. In fact, the market is far more spread out. Initially, Cogent did not have brand recognition or credibility on Waves. The fact that we have now sold Waves in 608 locations, we have 548 unique customers using Waves.

And in the last quarter, 3% of our Wave base or 77 waves upgraded their services. So I think we have crossed the credibility threshold. We have the footprint. At the end of the day, the customer buys based on value. Most investors hear the word value and they think price. But in a Wave, value is more multidimensional. It is the ubiquity of footprint, the speed to deliver, the diversity of the route, the reliability of the route and price. Each customer values one of or more of those 5 attributes different than others. Cogent has a superior value proposition compared to our national wave competitors in all 5 dimensions.

We have gone from being a non-player to 3% of the market in 6 quarters. We have an ultimate goal to get to 25% of the market. I think the successes we've had and the testimonials from customers will help us get there. Now there is a headwind, and that headwind is somewhat out of our control and not only affecting Cogent, but every company in the space, and that is supply chain constraints. So investors get excited when they hear companies say there's $7 trillion earmarked to go into AI training. That's a true statement. $1 trillion of that $7 trillion has been spent. $6 trillion is still to be spent. But that's created massive supply chain shocks.

Many data centers are announced or not yet completed. Many facilities need power upgrades. There's a shortage of GPUs. There's companies who build their whole business model on arbitraging GPU supply chain availability. There is a shortage of memory. I'm sure that was discussed at your keynote speech today with Marvell. There are shortages of routing equipment, server equipment, pluggable optics. For the first time in over 50 years, we are seeing technology prices rise. That does not mean Moore's Law is over. That just means this is a temporary aberration. These constraints have forced many customers to push out the acceptance of wavelengths.

We believe that many of these constraints will be ameliorated over the next year or 2 and the latent demand that we have been able to secure will then manifest itself in our revenue. And I'm sure Mike is going to ask me a question, can I predict what quarter that's going to happen in? And the answer is no. But I feel comfortable that the trends are here.

So, maybe going to the free cash flow side of the equation. So if we kind of count in CapEx or free cash flow to CapEx and the principal payments on your leases, what's the path to positive free cash flow and sustaining positive free cash flow for Cogent?

So there are 4 factors that impact that. The most important of those is growth in EBITDA. Our EBITDA actually accelerated due to the subsidy payments that we received from T-Mobile. When those payments stepped down, our EBITDA remained above what it was pre-acquisition, inclusive of the carrying cost of the empty Sprint network. On a going-forward basis, more of our EBITDA growth is going to come from high-margin revenue growth and less from cost savings. We have been able to expand margins and grow EBITDA. We know that in a little less than 2 years, that $100 million annual subsidy from T-Mobile will go away.

We've announced about $50 million in cost savings initiatives. That gets us half the way there and then the growth of our on-net services, we're quite comfortable that our EBITDA will exceed the EBITDA today in 2 years without the payments from T-Mobile.

The second impact is CapEx. Our capital expenditures pre-acquisition were running around $70 million. Sprint was spending about $30 million. That led us to forecast $100 million in CapEx. We knew that for a couple of year period, capital would surge as we reposition the data centers and reconfigure the network. We saw a material reduction in our capital intensity. In fact, our CapEx came down last quarter, $18 million sequentially. It is still running above the $100 million. It's running at about $140 million. We have experienced 7 price increases this year year-to-date from one vendor, 3 from another and 3 from another vendor.

We will asymptotically trend to that $100 million, but the path to get there will take longer. The third component has been principal payments on capital leases. There, we've been able to come in pretty much right on plan at about $10 million a quarter. We're paying down existing capital leases, but extending leases and signing new ones in new markets. Most significant new market for us has been India, but we'll continue to grow our footprint in the less developed world. We added Thailand recently to the network. We're doing some expansions in the Middle East. So there are new markets that are new revenue opportunities as the Internet becomes more globalized. It is not relevant to our Wave business, which is only U.S., Mexico and Canada.

And then finally, there is interest expense. We have paid down roughly $190 million of face debt. We bought back $138.9 million of our secured debt in the open market at a discount, and we also bought back $50 million of our unsecured debt, taking about $190 million off of the total debt load. Offsetting that, though, is the fact that our 2027 debt is maturing in June of '27. That debt is at a 7% coupon unsecured. The underlying base rates have gone up 150 basis points since we issued that debt and Cogent's aggregate leverage is up from 3.8x to 6.2x. That debt is going to be more expensive. We anticipate the increase in coupon effectively negates the savings that we got from gross reductions and that our cash interest load will be relatively equivalent to where it is today at about $180 million or $190 million.

Putting those 4 factors together and layering on top of that, the roughly $180 million that we still have due to us from T-Mobile and the cash that the company has on its balance sheet, we feel very comfortable in our capital and liquidity position.

Great. I guess to wrap up then, is there anything else you'd like investors to focus on as it relates to Cogent's story right now?

Yes, Roberta. Believe it or not, the Cogent story is a simple one. In many ways, it's probably the simplest company that you and Mike cover. But for whatever reason, we're shrouded in a veil of complexity. Our capital structure is complex. We're very transparent in our reporting metrics. In doing that level of transparency, we get investors who get fixated on a specific product or a specific market segment and then impute that across the entire business.

So when we see traffic growth grow at 15% year-over-year, that's very relevant to our transit business. It has no impact on our wavelength business. It has no real impact on our corporate business. But investors conflate these different levels of disclosure and oftentimes get confused. It really is a simple business with a fairly simple cash flow story. So my advice to investors is step back, look at the reported data, and see how well we have done over the past 21 years as a public company, and try to ignore the complexity and noise associated with the Sprint transaction, and you'll get much more comfortable with Cogent as an investment.

Dave, on behalf of both of us, thank you so much for joining us. It's great to see you, as always.

Thanks, Mike. Thanks, Roberta.

Thanks.

Good afternoon, everybody. Welcome to the Cogent Fireside Chat at the Goldman Sachs Communacopia and Technology Conference. It's my wonderful privilege to introduce Dave Schaeffer, who is the Founder and CEO of Cogent. My name is Mike Ng, and I cover telecom here at Goldman. We have about 35 minutes for today's presentation. First and foremost, thank you so much for coming out here to San Francisco and being with us this afternoon.

Mike, thanks for hosting me. I'd like to thank Goldman Sachs for a great venue, and I'd like to thank all the investors for taking time late in their busy day to hear what we have to say.

Wonderful. Well, I'm sure you will not disappoint. So Cogent operates one of the largest internet backbones in the world. Maybe talk a little bit about the current state of the business, your strategic priorities, how the company is positioned to benefit from all of these investments in AI infrastructure that we're hearing about this week during the conference.

So first of all, 84% of Cogent's revenues come from selling internet access or internet-based services. For 25 years, we were purely an internet service provider. Recently, we've expanded our business to include optical transport or wavelength services.

Let's start with the backdrop for internet traffic growth. And I'm going to bifurcate our market into two segments. I'm going to look at the market by end users. So today, about 2% of Cogent's traffic, but nearly half of its internet-based revenues come from selling to end users. Most of those end users are in large multi-tenant office buildings located in the central business districts of major cities. That business had historically grown at about 11% a year organically. When the pandemic hit, that growth rate declined to negative 9%. Today, it's back to being a 3% growing business. We today have about 35% market share in 1,870 multi-tenant office buildings where there is about 1.1 billion square feet of office space.

We also sell internet services to customers in buildings that are too small or too far from our network to justify us deploying capital. Most of those off-net customers actually came to Cogent through the acquisition of the Sprint Global Markets business from T-Mobile. We acquired a business that was declining and burning cash. T-Mobile paid us $700 million over a 54-month period. We still have just under two years of those payments left coming to us.

Our transit business, where we sell bulk internet connectivity in carrier-neutral data centers. The footprint is 94,000 route miles of intercity fiber connected to 33,000 route miles of metro fiber, serving 1,950 data centers in 58 countries around the world, approximately 7,500 access networks from companies as large as China Telecom or Jio by their upstream from us, as well as all of the major content producers. That business is a volatile business that is based on unit volume growth and price declines. Our transit business today is growing at about 8% in revenue. It is growing at about 15% in unit volume. We are the largest carrier of internet traffic in the world, carrying about a quarter of all internet traffic.

Roughly 77% of our traffic stays on net, meaning it goes from one paying customer to another. In 23% of the cases, we only get paid on one side, and the other side is a peer with whom we exchange traffic. This segment of our business has seen an acceleration as a result of AI for two discrete reasons. One, the traffic that is being collected is the raw material to build large language models and now is increasingly valuable to companies to transmit and collect for that model construction. Secondly, as we are transitioning to an inference phase, we have seen a significant acceleration in traffic growth due to agent-based or agentic AI driving traffic growth across the internet. We expect these trends to continue for multiple years.

The internet has gone through a series of discrete waves of growth. It had previously seen a major acceleration in traffic growth due to the acceleration in streaming at the beginning of the pandemic. That has since matured. We saw it grow slow and then re-accelerate now based on AI inference.

We have a couple of small bolt-on businesses. We lease IP address space out. We are the third largest owner of IP addresses. When you use the internet, you need an address just like you do to get mail delivered to your house. There are two competing technologies. The more desirable is IPv4. We have 38 million of those addresses. We lease out about 15 million of those addresses and generate about $70 million a year in that address leasing business.

We also have our own data center footprint. We operate about 170 data centers, about 1.4 million square feet, and about 150 megawatts of power. We recently divested of 10 unoccupied data centers for $225 million, selling those facilities to I Squared Capital. And we currently have 14 more facilities up for sale.

And then finally, the most controversial part of Cogent's business is its Wavelength business. This is selling point-to-point optical transport. It is a new business for Cogent. It was a business that did not exist inside of Sprint. We took the former Sprint long distance voice network that was built at a capital cost of $20.5 billion between 1982 and 1989. We purchased that asset from T-Mobile for $1 and then spent two years repurposing that to sell optical transport. Our transport business today is about a $60 million business. Last quarter, it grew 9.2% sequentially, about 62% on a year-over-year basis.

A major strategic initiative for Cogent is to grow any of our on-net services. The easiest of those is Wavelength. When we add $1 of on-net revenues, they carry 90% contribution margins. It is why over a 26-year history, we have averaged about 200 basis points a year of margin expansion. Our margin expansion took a material step back with the acquisition of the Sprint customer base. And as we have purged those less desirable revenues, we've resumed our margin expansion.

That's great. There's a lot there that I would love to go over over the next 25 minutes or so in the session. Maybe starting out with Waves, you know, a $60 million business, I think 3% market share relative to your long-term target of 25%. Talk a little bit about how Cogent's wavelengths maybe different or competitively different from those of your competitors. Thematically, what's driving increased demand for point-to-point optical transport? Is it speed? Is it privacy? Is it data center interconnect just as we have more distributed AI training? What do you see that's happening that's driving demand for Waves?

So it's all of the above, Mike. So first of all, the cheapest, easiest way to move bits is over the public internet. There is this thesis that the world converges to a single network. That, in fact, has happened. It's called the internet. It has cannibalized every other telecom service, and it has disadvantaged many legacy service providers. A wavelength, which is in fact a private line service, it is a point-to-point fixed connection, has three characteristics that cannot be replicated on the internet. The first and most important is defined latency. You know exactly how long it takes to get from point A to point B. The internet is a non-deterministic network, and by its very architecture is dynamically routing traffic over different paths to provide maximum resiliency.

The wavelength market or private line market has been around since the 1940s. It was initially delivered over fixed microwave by AT&T. MCI entered that market. In the late 1990s, the market pivoted to using optical transport. The term wavelength came into use. In fact, the first private lines that were sold on optical were dedicated wavelengths.

What is sold today is a subset of a wavelength. So we do not actually sell a full wavelength to any customer. We sell them a private line at either 10-gig, 100-gig, or 400-gig interfaces delivered over a wavelength. But it is a subset of the spectral capacity of that wavelength.

Initially, the market was dominated by government and large corporate users. The market really exploded with the advent of switch-based long-distance resellers, that market then collapsed. It was then replaced by an ISP market, which also collapsed. For the past 20 years, the market has had three primary use cases: international carrier network extension, regional carrier network aggregation, and content delivery. In the past three years, there is a fourth and significant new use case, AI training.

AI training requires wavelengths because approximately 60% of the capital cost of a training facility is in the GPUs. You want those GPUs running at 24/7 and not requiring buffering. The internet would slow down that GPU utilization, so companies generally pay about 2.5x as much per bit mile for optical transport as they do for internet services. It is also an unprotected product, which means that any physical point on the path can result in a failure if that point fails. That almost always requires customers to buy wavelength services in pairs, one is primary, one as backup. That market had become dominated by two major players, Lumen and Zayo. We had the opportunity in the Sprint acquisition to get paid to fix a broken enterprise business and to then be able to buy that $20 billion asset for $1.

An asset that doesn't produce revenues is not an asset, it's a liability. The carrying cost on that fiber was about $140 million a year. We continue to reduce that cost by selling off certain facilities, by optimizing the network, and the way in which we generate profit is by repurposing that fiber, by taking that 19,000 route miles of intercity fiber, connecting it to an additional 12,000 route miles that we've secured across North America, and then connecting it to about 22,000 route miles of North American metro fiber. We then interconnect that fiber to 1,137 data centers. We built a network that is truly unique, optimized for wavelengths, quick to deliver, can deliver any of three speeds, and deliver an any-to-any solution. So over 10 to the 3,000 power number of permutations. As we compete in this market, we have five discrete competitive advantages. We have more endpoints, faster delivery, unique routes, higher reliability, and lower price.

We have been selling wavelengths for about a year and a half. I wish we had sold them even faster and sooner, but we have gone from 0 to 3% market share. The uniqueness of our asset gives us a competitive advantage, and we've got the tailwind of this fourth use case in AI training that I think will allow us to reach our revenue target of $500 million in wavelength sales.

And on that wavelength revenue target, could you talk a little bit about the pace at which we'll get there? In the past, you've talked a little bit about data center construction delays as something that's affected customer acceptance. So where are we in that customer acceptance lifecycle? Are there any other delays that may be impacting wavelength acceptance?

So as of the end of Q2, we had installed and began recognizing revenue on about 2,500 unique wavelengths. Those wavelengths terminated as 608 unique data centers. They were sold to 548 unique customers. That is about one-tenth the number of waves that we ultimately need to sell to hit our revenue target. A wave is priced in a three-dimensional grid. It looks at the length of the path, the size of the wave, and the duration of the contract. We continue to see a migration to longer wavelengths and a migration to higher throughput.

What has been frustrating to us is that some demand is not being fulfilled because customers, while they want the service, cannot accept the service due to limitations in their architecture. Those limitations could include the fact that a data center is not built, a data center has insufficient power, perhaps there is insufficient server capacity, and sufficient memory, and sufficient GPUs, and sufficient pluggable optics, perhaps insufficient router equipment for those facilities.

We have focused on the $7 trillion of announced AI training capital that is being spent. About $1 trillion of that $7 trillion has been spent. $6 trillion is yet to be spent. Also, it is not a linear expenditure in the sense that there are certain items in the supply chain that may be constrained, and most of the capital is spent, but not all of it until all of those items are delivered. And it has frustrated us. I think we will continue to see a gain in market share and a willingness of customers to broaden their view of Cogent from being just an internet provider to now being also a provider of optical transport.

I want to touch on the data center sales. So, as you mentioned, 10 data centers sold. What was the price per megawatt on those data centers? And then, do you expect the remaining 14 facilities that are for sale to transact at a similar price?

So we sold a total of 54 megawatts across 10 facilities, approximately 500,000 square feet at $4.3 million per megawatt. We have 14 additional data centers up for sale. We have LOIs on four of those additional. The pricing is similar. Each data center is slightly different based on its geography, its square footage, its existing power, and its potential for power upgrades. All of these facilities were originally built as telephone switches.

When we first acquired Sprint, we did not intend to invest in the switch footprint and rather focus almost exclusively on the wavelength business. About a year after we announced the acquisition, we had a number of inbound inquiries, and it became clear to us that the 230 megawatts of power that was provisioned for those phone switches had economic value.

We evaluated all 482 facilities. We chose 125 to convert. We earmarked $100 million of capital that we did not have. It increased our leverage, and between June of '24 and June of '25, we did the repositioning of these former telephone central offices into data centers. We then placed them in the market for sale. We quickly signed some letters of intent, and we closed our first sale on June 29, 2026, resulting in 10 facilities and $225 million in gross proceeds.

Because of the tax attributes that we had and the fairly de minimis transaction costs, we will recognize about $224 million as a tax gain, but are able to shelter that and should be able to retain virtually all of that gain. We then used a large portion of that, $175 million of the $224 million, to repurchase debt at a discount in the open market. We repurchased $138.9 million of our 2032 secured debt at an average of about $0.90 on the dollar. And then we also repurchased about $50 million of our unsecured debt at a slight discount of about $0.99 on the dollar.

I wanted to ask about Cogent's margin outlook. It's something that Cogent has been able to address driving margin expansion because of the wind down of the Sprint integration cost. Now that the bulk of the work is behind you, what cost efficiency initiatives are you prioritizing next? How are you thinking about margins going forward?

Yes. So let's maybe go back and look at Cogent's margin trajectory pre-Sprint. We went public in June 2005 with 0% EBITDA margin. So it was our first quarter of turning EBITDA break-even. Over the next 18-year period, we averaged 220 basis points a year of margin expansion. That margin expansion came from the fact that roughly 3/4 of our sales were on net, 1/4 of our sales were off net. Those on-net sales carried 90-plus percent contribution margins and allowed our margins to grow from 0 to 40.5%. In May 2023, we completed the acquisition of the Sprint operating business from T-Mobile. That business had negative 60% EBITDA margins, was declining at 10.9% a year, and represented 42% of the revenues of the combined company.

In the first quarter, post-transaction, our EBITDA margins fell from 40.5% to below 2%. Our annual EBITDA was running at only $5 million. Now, to offset that, we got $700 million, $350 million of it in the first 12 months, the other $350 million spread over the next 42 months in the form of a subsidy payment from T-Mobile. With that added in, our margins actually expanded all the way up to 47% initially and then receded back to about 15%.

The underlying margins without the subsidy payments have gone from a 2% margin to about a 20% margin today. We anticipate being able to grow our margins annually and at least 200 basis points a year until we resume margins in the mid-40s. So we have a long runway ahead of us. That margin expansion comes from three primary drivers: cost reductions, product rotation, and new product sales primarily of on-net services. Over the past three years since closing the Sprint transaction, our aggregate reported revenue went from being positive 10.2% to negative 5%.

We've had 12 out of 12 quarters of negative revenue growth. Yet 11 of those 12 quarters, we've actually delivered EBITDA growth, not just margin growth, but absolute EBITDA growth. That should continue. We still have about $8 million of cost savings as of the end of the quarter -- end of Q2, to achieve. We also will achieve about a $7 million annualized savings from the divestiture of the data centers. And we are eliminating $36 million in expenses associated with integration. This is above and beyond the contribution margins from on-net sales.

In the most recent quarter, Q2 of 2026, our on-net sales were 82% of sales. Cogent's revenues actually have grown every quarter since the deal has closed by approximately 29% or about 8% per year. That's been masked by the fact that the acquired Sprint revenues have declined by 69% during that same period, resulting in the total top line reported revenue trajectory of negative 5%. But on a going forward basis, we should see more of our EBITDA growth come from top line growth because the acquired Sprint business has become only about 15% of the combined company's revenue as opposed to being 42% at closing.

Great. I wanted to get one in on capital allocation. Where are you in terms of refinancings? How do you think about further debt reduction, network reinvestment, and then restoring capital returns to shareholders?

So to remind investors, Cogent has returned approximately $2 billion to equity. Virtually all of that has been tax deferred. Our dividends, which have been treated as a return of capital, were about $1.7 billion and about $300 million in the form of share repurchases. We have dramatically reduced that capital return program in order to delever. The acquisition of Sprint increased our net leverage from 4.2x to 6.7x. Today, we're at 6.23x net leverage as of the end of the quarter. We have a stated policy of getting back to 4x net leverage before we will meaningfully return capital to equity. That return of capital will probably be a combination of dividends, hopefully treated as return of capital, and also in the form of continued buybacks.

In order to provide the company adequate liquidity, we have been repurchasing some debt at a discount. We also know that we have $700 million of our unsecured debt that comes due in June of '27. In order to address that maturity, we are working on converting the unsecured debt to secured debt. We have ample capacity within our indentures to do that. And then using that secured capacity to do a new $700 million secured offering. We have been working with another bank on this. There is not yet a formal transaction launched. We have been in discussions with both existing bondholders and potential new bondholders and are looking to optimize our cost of capital.

Great. And then in the last minute that we have, I would just love for you to talk a little bit about key things that you think investors should be focused on as it relates to the Cogent story.

Yes, listen, it's been a tough year for Cogent's equity. That equity has sold off meaningfully. The fundamentals of the business are actually stronger today than they've ever been. Those fundamentals are somewhat masked by the complexity of a transaction where you get $20 billion of asset value for $1. You get an operating business doing $485 million in revenue with negative $300 million of EBITDA, and you're paid $700 million to take it. And I think it is that accounting complexity, the difficulty investors have in parsing out the underlying Cogent story from the totality of the combined company that has penalized us and raised our cost of capital. We firmly believe our cost of capital will come down and we'll be able to continue our track record of returning capital to equity.

Well, Dave, thank you so much for participating in our conference. It's been a privilege to have you on stage here.

Hey, thanks, Mike. Thank you all very much.

Good morning and welcome to the Cogent Communications Holdings second quarter 2026 earnings conference call As a reminder, this conference call is being recorded and it will be available for replay at www.cogentco.com. The transcript of this conference call will be posted on Cogent's website when it comes available. Cogent's summary of financial and operational results attached to its press release can be downloaded from the Cogent website. I would now like to turn it over to Mr. Dave Schaffer, Chairman and Chief Executive Officer of Cogent Communications Holdings. You may begin.

Hey, thank you and good morning. Welcome to our second quarter 2026. conference call. I'm Dave Schaefer, Cochran's Chief Executive Officer, and with me on this morning's call is Tad Weed, our Chief Financial Officer. I'd like to focus on a few key events and significant matters that transpired in the quarter. I'd like to recognize these events and give you an update on these important matters. We have made significant progress in several areas, our data center monetization, our net leverage reduction or cost reduction and completion of various integration projects The continued product rotation into more profitable on-net services a reduction in our capital expenditures and a reduction in our capital lease payments, and continued progress in the sale of wavelength services. First, for data centers and leverage.

As we stated in our previous call, we intend to monetize 24 of the facilities that we acquired from the Sprint acquisition and converted into data centers, either through the outright sale or leasing these facilities on AHO sale basis. basis in june we closed on the sale of 10 of these former sprint facilities that we converted into data centers for total proceeds of 225 million paid in cash by the purchaser in the quarter The sale of these assets resulted in a gap gain of $130.7 million. We intend to use the majority of these proceeds from the transaction to reduce both our gross leverage and our net leverage. We reduce our net leverage as adjusted inclusive of our payments from T-Mobile in this quarter to 6.23 times EBITDA from 6.79 at the close of last quarter and from 6.25 to 6.25. 6.61 times in Q2 of 2025. We continue to have multiple parties interested in the remaining former Sprint facilities that that we have put up for sale. We are in negotiation for several letters of intent on these facilities. Our total cash and restricted cash balances at the end of Q2 2026 was $369.7 million. Now I'd like to touch on the amendment that we received to our 2032. secured note indenture.

In June, we obtained approval from the majority of the holders of these 2032 notes to amend the note with a supplemental indenture. The key features included in this revision are an increase in maximum secured debt leverage ratio, from four times to 4.7 times, 475 times. A commitment on our part to use at least $175 million in proceeds from the sale of these initial data centers that was then contributed from outside of the borrowing group into the borrowing group to be used for the repurchase of debt obligations at a discount. During the quarter, we repurchased $20.4 million of par value 2032 notes at an average price of $91.955, resulting in a gain in the quarter of $1.6 million. $1.4 billion. Continuing in July, we purchased an additional $118.4 million of par value notes at an average price of $90.071, resulting in an additional of $11.8 million, which will be recognized in the third quarter. The total purchases of our 2032 notes to date have been $138.8 million of face value, at an average price of 90.348, resulting in a cumulative gain of 13.4 million. We are making progress on the refinancing of our $750 million 2027 unsecured news.

Our $750 million unsecured notes mature in June of 2027. The May cold period for our 2027 unsecured notes ended on June 15, 2026. These notes have become current, and we are in the process of completing or refinancing of these notes. We expect that transaction to be completed. to be completed in the third quarter of 2026. Now for a couple of comments on our wavelength business. Wavelength business continues to grow. At quarter's end, we are offering wavelengths in 1,137 locations. with 10 gig, 100 gig, and 400 gig services available and provisioning intervals in approximately 30 days, which do continue to improve. or length revenue for the quarter was 14.8 million, an increase of 63.8% from a year ago, and a sequential increase of 9.2%. Our wavelength customers increased year over year by 66.4% and sequentially by 8% to a total of 2,415. 45 customer connections.

In addition, during the quarter, to the new installs that we have reported, we re-provisioned 77 existing wavelengths, converting them into higher capacity wavelengths. Most of these were conversions from 100 gig to 45 gig. 400 gig waves as customers have become more confident in the quality of our network. At quarter's end, we have sold wavelength services in 608 unique locations, and we We have sold those wavelength services now to a combined customer base of 546 unique customers. We still believe that we will capture 25% of the North American long haul wavelength market. We also today still have only captured 3% of that market. Now for a comment on our gross margin improvement. We continue to reduce costs. Our gross margins percentages increased on a year-over-year basis by 260 basis points and increased sequentially by 90 basis points to 47%.

Our EBITDA, EBITDA as adjusted and EBITDA adjusted margins also improved. We expanded our sequential EBITDA as adjusted margin. Our EBITDA as adjusted for the quarter increased sequentially by 9.2%. $900,000 or just under $1 million to $71.1 million. And our EBITDA as adjusted margin increased sequentially by 90 basis points to 30.2%. We also have worked diligently on the organizational optimization of our workforce. As we are completing various integration projects, we are evaluating the optimal size of all of our departments as the integration of these former Sprint assets into Cogent is now being completed. We reduced our total headcount to 1,682 at quarters end, a reduction of 113 individuals from the end of the previous quarter, and a reduction of 207 individuals from Q2 of 2025.

This reduction represents approximately 6% of our workforce from the previous quarter. The expenses associated with these reductions have been recognized in the second quarter. Now I'd like to take a moment to talk about our long-term objectives and beliefs around targets. We expect our revenues to grow at between 6 and 8% over a multi-year period. While we acknowledge our revenue growth in Q2 of 2026 was negative, we do believe that the decline in revenue from the acquired Sprint customer base is moderating. anticipate EBITDA margins to average over a multi-year period approximately 200 basis points a year, kind of mirroring the type of margin expansion that Cochran had experienced prior to the acquisition of Sprint. Our revenue and EBITDA guidances are not intended to be quarterly or targeted to a specific year, but rather a multi-year. Now I'd like to ask Tad to read our safe harbor language and provide some additional details on our operating performance for the quarter.

I'll then conclude with a few summary remarks, and we'll then open the floor for questions.

Yes, thank you Dave. Good morning everyone. This earnings conference call includes forward looking statements. These forward looking statements forward-looking statements are based upon our current intent, belief, and expectations. These forward-looking statements and all other statements that may be made on this call that are not historical facts are subject to a number of risks and uncertainties and actual results may differ materially. Please refer to our SEC filings for more information on the factors that could cause actual results to differ. COGEN undertakes no obligation to update or revise our forward-looking statements. If we use non-GAAP financial measures during this call, you will find these reconciled to the corresponding GAAP measurements in our earnings releases that are posted on our website at cogenco.com. discussion of the results for the corner and the revenue mix since spring closing which the full course quarter was Q3 23 versus this quarter Despite our revenue decreases, we have been able to increase our margins.

Our increases in our gross margin and our EBITDA margin have been driven by cost reductions and a rotation to our more profitable on-net products. Comparing our revenue by connection type from the third quarter of 2023, which again was the first full quarter we were combined with Sprint Wireline, to this quarter illustrates the material changes to the composition of our revenues and the strength of the underlying criminal classic business. Our on-net revenues were 47% of our total revenues in the third quarter of 2023. Our total on-net revenues, including on-net wavelengths, increased 63%. So close to 64% of our total revenues this quarter, and that was compared to 62.4% last quarter and 57.4% in the second quarter of last year. Our off-net revenues were 48% of our total revenues in the third quarter of 23, are much less profitable. Our off-net revenues have decreased to 35.9% of our total revenues this quarter, compared to 37.2% last quarter and 41.5% in the second half. quarter of last year. 18% of our sales this quarter were for on-net services, In the aggregate, our non core revenues were 5% of our total revenues in the third quarter of 23.

And they have decreased to less than a 1M dollars and we're about 0.4% of our revenues this quarter. Our total revenues for the quarter were $235.6 million. revenue declined by 3.6 million to 1.5%. USF tax revenues had a negative impact on our sequential revenue results of 0.6 million and a negative year-to-year impact of 1.1 million. The combined impact of USF tax and FX had a negative impact combined of $0.8 million on our sequential revenue results. We analyze and classify our revenues into four network connection types and three customer types. Our four network connection types are on-net, off-net, wavelength, and non-core. Our three customer types are net-centric, corporate, and enterprise customer.

For the quarter, sequentially, our on-net revenues, including wave revenues, increased by $1 million. Are less profitable off net revenues declined by four and a half million, so most of the decline was related to off net. non-core revenues decreased by 0.1 million. Our wavelength revenues by themselves, which is almost entirely on that, increased by 1.2 million. IPV4 lease revenue, which is included in OnNet, our OnNet IPV4 leasing revenue increased sequentially by a half percent to $18.1 million and 18.1% year over year. our lease price per address has been stable for the last several quarters and was zero 40 cents per month. We have titled the 37.8 million IPv4 addresses and we've leased approximately 15 million IPv4 addresses as of today. The substantial changes in the acquired Sprint Wireline revenue base have masked the underlying performance of our code and classic business. Our consolidated revenue declines have been largely attributed to the reduction in the acquired Sprint Wireline corporate and enterprise non-core and off-net revenues.

At closing, the Sprint Wireline revenues were 42% of our total revenue. That has declined to only 15% of our revenues this quarter. We acquired the Sprint wireline with a revenue run rate of $118.25. This acquired revenue base has decreased from 118 million and down to 34 million for this quarter. That's an $84 million reduction in quarterly revenues related to our acquired SprintWireline revenue base, or a 71% decline since deal closing. At deal closing, which was three years ago, our Cogent Classic revenue run rate was $155 million per quarter, and And the coaching classic revenue base has increased from then by 29% from $155 million to $200 million for this quarter. Revenue by corporate, net centric, and enterprise.

Our total corporate business represented 41.9% of our revenues for the quarter. That decreased by 9.6% year over year and sequentially by 2.4%. Our total net-centric business continues to increase and to benefit from the growth in video traffic, activity related to artificial intelligence, streaming, ICD-4 leasing, and wavelength sales. Our net-centric business represents 45.6% of our revenues this quarter. Our quarterly net-centric revenues increased by 10.4% year-over-year and sequentially by 1.6%. Lastly, our enterprise business, our total enterprise business was 12.5% of our revenues this quarter. Our quarterly enterprise revenue decreased by 26% year over year and sequentially by 8.9%. primarily due to reduction in the acquired Sprint, Wireline, Enterprise, off-net revenues as non-core is down to less. less than a million dollars.

Revenue and customer connections by network type. We serve our on-net customers in 3,627 total on-net buildings. Our total on-net revenue, including on-net wavelength sales, was $150.2 million for the quarter, a year-over-year increase of 6.2% and a sequential increase of 0.7%. Our off net revenue was 84.5 million for the quarter. year-over-year decrease of 17.3% and a sequential decrease of 5.1%. Our off-net revenue results are impacted by the continued grooming and termination of low-margin off-net contracts, and particularly the acquired for wireline customers. From pricing, our average price per megabit for installed base decreased sequentially slightly to 11 cents from 12 cents last quarter and from 17 cents for the second quarter of last year. Our average price per megabit for our new customer contracts also slightly declined with $0.06 compared to $0.07 last quarter and $0.08 in the second quarter of last year.

OUR ARPU'S FOR THE QUARTER WERE AS FOLLOWED. OUR ONNET IP ARPU WAS 513. off net IPR was 1197. Our wavelength ARPU was $2,100. And our wavelength ARPU for new waves this quarter was 2,206, as there were more larger connections installed. Our IPv4 ARPU, again, was 40 cents per address, very stable. Our churn rates, our on-net churn rate, monthly churn rate slightly increased to 1.3% from 1.2% last quarter. our off-net churn rate is primarily driven by the reduction in the acquired sprint customer base. And that rate was 2.3%, an increase from 1.7% last quarter. Lastly, our wavelength monthly turn rate was about 0.5%.

Traffic. Our IP network traffic growth continued for the quarter. Our IP network traffic growth for the quarter increased sequentially by 3%. In what is a traditionally seasonally slow quarter for traffic growth and year over year at an accelerated rate up to 16%. Sales rep productivity, our sales rep productivity materially improved substantially and was 4.5 this quarter compared to 4.1 last quarter. Our long-term average is 4.8. Comments on FX, our revenue earned outside of the United States was about 21% of our revenues for the quarter, very consistent. Based on the average Euro and Canadian conversion rate so far this quarter, so in the third quarter, we estimate that the FX conversion impact on sequential revenues will be negative and 0.3 million and year over year also negative 0.8 million. Customer concentration, our revenues and customer base are not highly concentrated.

Our top 25 customers are 16% of our revenues in support. CapEx and payments on capital leases. Our CapEx declined by 16.7% sequentially and 31.4% year-over-year to $38.5 million for the quarter. We continue to experience multiple equipment price increases from vendors due to supply chain constraints so far this year. Our principal payments on capital leases also declined sequentially by 27.7% and were $9.7 million for the quarter. Debt and debt ratios. Our total gross debt at par, including our $630.2 million of finance IRU leases, and our reduced principal amount of our 2032 notes, which a quarter round was reduced from $600 million to $579.6 million. The total was $2.2 million. $2.3 billion and a quarter in.

Our net debt, total debt net of our cash, and our 151.5 million amounts due from T-Mobile was 1.8 million. The principal balance on our 2032 notes again has been reduced further after quarter end and is now 461.2 million from the purchases we made in July. Our leverage ratio as calculated under our more restrictive unsecured 750 million 2027 notes was Our secured leverage ratio under the notes was 3.67. Our fixed coverage ratio was 2.28. The definition of consolidated cash flow under our 600 million year 2032 notes and venture includes cash payments under our IP transit service agreement with T mobile in the determination of consolidated cash flow under the indenture. And those ratios were as follows. Our leverage ratio as calculated under the $600 million note indenture was 4.56, secured leverage was 2.81, and fixed coverage was 2.97. cash and restricted cash. 168 million of the proceeds from the sale of the 10 data centers was considered restricted cash at closing of the terms of our supplemental indenture since that amount was reserved for purchases of our debt obligations at a discount.

We purchased 20.4 million par value, or 2032 notes, in June. As a result, the balance of restricted cash related to the data center sale proceeds was $147.6 million as of June 30. Again, we purchased an additional 118.4 million par value of our 2032 notes in July, so the remaining balance of the restricted cash, 29.2 million, as of July 31st. The cumulative purchases of our 2032 notes were $138.8 million of par value. That was retired for paying $126.2 million at an average price of $90.348 and the cumulative total gain $13.4 million. Lastly, bad debt and day sales. Our DSO improved and was 29 days at quarter end, a two day improvement from from 31 last quarter and our bad debt expense was only 0.6% of our revenues for the quarter.

And with that, I will turn the call back over to Dave. Hey, thanks, Ted. I'd like to highlight a couple of strengths around our network, our customer base, and our sales force. We remain direct beneficiaries of increased traffic volumes from over-the-top video, artificial intelligence activity, streaming and gaming trends. At quarter's end, we were able to sell WAVE services into 1,137 unique data centers with reduced provisioning windows for approximately 30 days. We sell those WAVES in 608 of those locations as of today to 546 unique customers. At quarter's end, we were signed for IP services globally in 1,953 data centers. At quarter's end, our IP network remains the most connected in the world with 7,000 7,572 networks directly connected. 22 of these networks represent peers and 7,550 networks are paying cogent transit customers.

We remained focused on our sales force productivity and continued to manage out underperforming reps. Our sales force turnover was 7.5% a month in the quarter, which is above our historical average of 5.7% per month. At quarter's end, we had a total quota bearing sales force of 506, 263 of these professionals focus on the net centric market, 230 focus on the corporate market, and 13 focus on the enterprise market. We have made significant progress in several areas. We continue to improve our margins, grow our EBITDA due to our diligence and cost reductions, the completion of many of the integration programs that we outlined three years ago, focus and selling more profitable on-net services. In the second quarter, 82% of all sales in the quarter were on Net Services. And as a result in our revenue base, the percentage of revenues that come from on net increase sequentially to 64%.

We're actively working on further monetization of the converted Sprint facilities. ability to increase EBITDA margins through optimization and asset divestiture will accelerate our delivering and allow us to resume a program of aggressively returning capital to equity holders. We remain disciplined and our capital investments were focused on expansion projects with the highest return on incremental capital. or wavelength services are differentiated by the quality, breadth, of our footprint, uniqueness of our routes, and efficient provisioning capabilities. Our on-net services, whether they be WAVE or IP, are unparalleled in their value to customers. At Quarters End, we're providing services in 1,781 carrier-neutral data centers and 172 cogent data centers. This footprint and aggregate reaches facilities... with approximately 17 gigawatts of installed power. The cogent data centers that we operate have a total of 155 megawatts of installed and available power and over $1.5 million. square feet. The proceeds that we have been able to garner from these data center sales have allowed us to reduce leverage and has allowed us to add resources to the marketing of the remaining facilities.

We are in the process of completing the refinancing of our 2027 notes, which we anticipate will complete in the third quarter of this year. We offer superior products, unparalleled quality, broad footprint into traffic locations with expedited provisioning and disruptive pricing. In summary, we continue to gain market share by the value we deliver our With that, I'd like to open the floor for questions.

Thank you. We will now begin the question and answer session. If you have dialed in and would like to ask a question, please press the star 1 on your telephone keypad. And if you would like to withdraw your question, please press the star 1 again. Your first question comes from the line of Greg Williams with TD Cowen.

Great, thanks for taking my question. First one's just on waves. Your peers over the last two weeks noted some strong wave numbers. Your numbers came in a touch light to the estimates, mine and the streets, I think. And I realize you don't provide the backlog KPIs, but any help on backlog direction would be helpful. As we're wondering here, is this still about customer service? delaying or not accepting orders. And you reiterated the 25% share in the long term. Like what needs to happen? You're a far cry from that number. as you think about the timing of that target now.

Second question is on EBITDA. It grew quarter-over-quarter as message, but it barely did so. one-time cost you mentioned head count reduction how much was that and is that in your EBITDA and are we fully done with the cost takeout progress essentially we're just trying to figure out what the EBITDA cadence looks like for the balance of the year thanks.

Yes, sure. Thanks for the questions, Greg. With regard to wave sales, our demand remains strong. We continue to add to our backlog. You are correct, we are no longer disclosing that backlog specifically, we also are encouraged by the fact that customers that have used our services are now coming back and asking to increase the capacity on those wavelengths and helping us push ARPU up. We have been frustrated by the fact that many customers struggle to have the ability to use the waves that they've ordered, whether it be equipment deliveries, power constraints, or data center space. space and cooling availability or even data center completion. So because of the surge in demand for computing, the entire supply chain is adjusting, and that has impacted our customers. It has not to date slowed our ability to provision, but it has increased our capital.

And you know, the fact that we were able to have a meaningful reduction in our capital spending sequentially and and expect that number to continue to improve is because of our efficiency and deploying that capital. You know, in terms of wavelength market share, you know, we are at only 3% of the North American long haul market today. You know, we remain encouraged by the breadth of customer base that we have. 546 unique companies using our wavelengths, getting comfortable with Cogent, and the fact that we've now delivered those into more than half of our wave enabled footprint, 608 out of the 1137 wave enabled locations. You know, I think we expect our rate of wave installs to accelerate. But I do think it will be several years till we get to that 25% market share. I'm going to now pivot over to the EBITDA progression and. And. You know, while we experienced a significant revenue decline of $3.6 million sequentially, our EBITDA did grow by nearly a million dollars on a sequential basis. basis embedded in those costs and retarding our rate of margin expansion were some of the expenses that we have incurred in the quarter as we have wound down many of the integration projects just to remind investors.

When we initially acquired Sprint, we had targeted $220 million of targeted direct cost savings and we a monthly integration expense of about $5 million a month. We updated those numbers and as recently as last earnings call, we had taken that $220 million number to 240 and we had indicated that of that 240, there was less than $20 million of annual the annualized run rate remaining. We have brought that number down. There is still a small stub of remaining costs, but the vast majority of the 240 million in cost savings have been achieved. We also commented over the last three years on the decline in our spending on integration projects. You know, that monthly spend of $5 million a month had declined to about $3 million a month at the end of Q1. We accelerated that decline in large part through the optimization of our workforce.

While we had been gradually reducing our workforce, that rate of reduction accelerated in a quarter and we reduced our workforce by about 6%. There will be further reductions in the third quarter, but at a more moderate rate. The cost of these reductions, severance, benefits, payments, and the fact that many of these employees did continue to work at least partially in a quarter did impact the rate of margin expansion. I do believe that over the next several quarters, we will return to a more accelerated rate of margin expansion through the combination of continued on net sales and the continued discipline around taking out those integration expenses. As we had outlined previously, going back to September of 2022 when the deal was announced, that we anticipated all of those integration costs to be gone by the end of 2026, year end. We will probably be in a position to beat that, but we have materially reduced those expenses, and therefore we'll see uplift in both the third and fourth quarter from that.

Thanks. We have a quick follow-up. You mentioned that CapEx was down and you expect that number to continue to come down. Is that a quarter-over-quarter or near-term target? Is that sort of a longer-term view?.

So it was both on a year over year basis and sequentially that capital declined. We would expect a further decline on both a sequential and year over year basis in the third quarter. You know, as I commented extensively on the last call, it has been difficult for us to give exact capex guidance because of the pacing of equipment price increases. We had experienced five increases from one vendor in the first five months of the year. Fortunately, in the next three months, we've only experienced one increase, and they were relatively equal size. Our other primary vendor has had three increases, and we do believe that that rate of price increase is moderating, and we do expect our total CapEx spend to continue to moderate.

Got it, thank you. Hey, thanks, Craig. And your next question comes from the line of Chris Scholl with UBS. Please go ahead.

Great, thank you. Dave, just to follow up on the status of those 14 Sprint data center sites that have been converted, do you envision selling those 14 as a block or as a piecemeal approach more likely based on the interest you're seeing so far and any color you can give on the number or types of buyers expressing interest at this stage and what the timeline looks like?.

from here. Thank you. Yes, sure, Chris. I think it's most likely that they will be sold in chunks. We do have numerous both operators and private equity looking at these facilities. We currently have signed letters of from purchasers that we have not accepted that we have validated their credit worthiness and those are for four facilities two letters of intent one for three one for one facility and You know, we are also in discussion for many of the remaining 10 facilities, but we do not today possess letters of intent. The letters of intent that we have, Cujon found unacceptable in terms of price based on the characteristics of these facilities, and we are in the process in negotiation. That is not a guarantee we'll get a deal done with those counterparties, but we remain optimistic. I think for the other 10 facilities, the counterparties are still being vetted and We also do not yet have a firm offer from them, in some cases verbal indications of interest, but I think it's our intention to only announce once we have something that we feel will absolutely close.

And we're encouraged. by, you know, I squares progress and their professionalism and moving from letter of intent to contract to closing.

If I can just follow up on the 10 sites you did sell, can you just remind us how much expense should drop out in 3Q for those assets, and is it fair most of that spend is in COGS.

Yes, so the expense associated with those facilities on an annual basis was about $7,000,000. So we would anticipate just under $2,000,000 of direct expense comes out. Offsetting that is we remain a tenant and a small footprint in those facilities for a few hundred thousand dollars of expense for space and power for our equipment that we intend, just as we're in other data centers. to continue to operate and in addition to those savings we also will be receiving payments from i-squared for providing transition services to them that will be recognized either as an offset to cost or revenue. Great. Thank you.

Hey, thanks, Chris. And your next question comes from the line of Mike Funk with Bank of America. Please go ahead.

Good morning, Dave. How are you? Hey, great, Michael. Yes, good to hear from you again. So, one of the back to wavelengths, Dave. We really have not seen acceleration and connections, a lot of us had anticipated. So, can you maybe just talk a little bit about the conversion of some of that backlog, and what the impediments have been. And then on the sale or potential sales or remaining Sprint data centers, love to get your thoughts on how we should think about valuation, based on the location, quality, tenants, whether that would be comparable to valuation the data center has already sold,.

you closed on? Michael Sink – yes, sure. Two very good questions, Michael. So first of all, on wavelengths, you know, the constraints I think have fallen into two primary categories. One, customers not being able to accept the wavelengths because of constraints that they are facing. You know, existing data center occupancy is at a record high. Many data centers do not have surplus power available so the customer can't put the equipment in. to accept the wavelength. There can be other supply chain constraints around, you know, whether it be servers, routers, switches, pluggable optics, all of these components that allow the customer to use the wavelength that we deliver have elongated delivery times. I think secondly, and we commented on this several quarters ago and have talked about repeatedly, customers, I think still are amazed at our ability to provision as compared to others you know i know the question and comment came up about others reporting quote-unquote strong wavelength sales to the best of my knowledge none of our competitors give the level of granularity on wavelength sales that we disclose the number of wavelengths the ARPU, and they oftentimes do not break wavelength revenue out as a separate line item.

So, you know, it's one thing to qualitatively say things are great. That's another to be very granular and specific. I'm not saying that in a defensive way, but I do think our transparency on this has allowed investors more clarity and more granularity than our competitors. You know, we today have equipment and are able to continue to deploy wavelengths. We have had to make decisions around, is the customer who has an order really going to take the wavelength? And then, two, prioritize some. equipment and resources to customers that already are using wavelengths and are upgrading. You know, I think we did better than the headline number, you know, demonstrates in the quarter with the re-provisioning of 77 wavelengths on top of the 182 that we deployed. These are still relatively small, numbers.

You know, we have a strong year over year growth rate, you know, at over 60% percent, a strong sequential growth rate of 10 percent. These are, I think, admirable, but we need to continue to grow the base. We do believe that the demand is real, the customer base is broad, the footprint we have is the footprint that customers want. There's been conversations around Cogent's willingness to go either into proprietary single-tenant data centers or to corporate sites. We do accommodate waves to those locations. We have provisioned them, but we do so in a different manner. We we usually look for the single tenant at that data center to provide dark fiber extensions back to a carrier neutral, where we then interconnect and hand the wavelength off.

This allows us to minimize our capital exposure, yet still meet the requirements of the customer. That is primarily the hyperscalers. And then for large enterprises, where we have sold a handful of waves, they are typically buying them in single tenant office buildings. And there we use a combination of Mark Fiber Tales and local providers to interconnect because we cannot justify the deployment of capital for that single tenant opportunity. I'm going to now pivot over to your question around the complexion of the unsold data centers, both in terms of quality and anticipated price. The facilities that we have sold are, I think, you know, comparable to the facilities that we have to sell. There is clearly a divergence in scale from the largest of the remaining 14 being our Fort Worth facility with 14 megawatts and the smallest of the remaining to be sold, having just less of one megawatt in Pearl City, Hawaii. You know, I think some of the remaining sites have incremental power available above and beyond what we have today provisioned.

We've got written in confirmation from utilities that there is extra power and we place some value not the same value as fully provisioned power, but that is part of the back and forth on our negotiations with the potential buyers. You know, the geographic footprint remains diverse, and I do believe that, that most of the remaining facilities will eventually transact. You know, it's really only been. a few quarters since these facilities were fully converted. And I think, you know, there are probably some private equity investors waiting to see kind of high squares business strategy and potentially, you know, replicate So I think over the next several quarters we will be able to transact or more. The final point is, you know, based on Cochran's North American NOL inventory, I think it may be optimal for us not to transact in North America until early next year when we will have some additional NOL capacity to offset taxes.

Understood. Just really quickly, can you remind us please the total megawatts in the 14 remaining facilities?.

Yes, so the price per megawatt that we transacted with I2 was approximately $4.2 million a megawatt, and that inventory was 55 megawatts, roughly, and then the remaining footprint, is about 55 megawatts. Great. Thank you, Dave. Hey, thanks, Michael.

And your next question comes from the line of Walter Pysak with LightShed. Please go ahead.

Hey, Dave, I want to go back to the first question. You give a very comprehensive answer, but I just want to dissect it a little bit to understand it. On SG&A specifically, as it relates to... synergies and then ongoing integration expenses. I know in Q1 you have your typical sales meeting. So I thought with, you know, with basically the ongoing synergy stuff, you might see a decline in Q2 that didn't happen. But I think I think you said in that first answer that's because of ongoing integration expenses. I know you were talking a little bit bit more about gross margin, but I guess if we could just focus on SG&A, should that SG&A or is there more room for that SG&A to decline on an absolute basis as we conclude the rest of the year?.

The answer directly to your question is yes, it will decline sequentially into the third quarter and the fourth quarter. Embedded in that SG&A number, in addition to the sales meeting in the first quarter, were, you know, the auto. the credit expenses associated with the first quarter and the increase in employee load due to FICA matches, which do max out and typically go down. But in addition, in the second quarter. We had a accelerated rate at which we ended integration projects. There are still some ongoing, but at a much reduced rate. And we took out 6% of our sales force rather than many companies which put about a press release of what they're intending to do, and then the expenses followed, we took a slightly different approach and took those employees out on a very tactical basis that were related to many of these integration projects. There were severance claims. costs. Those terminations did not all occur early in the quarter, and we would expect to see a flow through in SG&A improvement in both Q3 and Q4.

Yes, that makes sense. And then on the IP addresses, growth seems to have stalled out a bit. So I guess it goes back to the age-old question. kind of given the debt leverage, I know you just kind of detailed what you hope for in the data centers, but like, why not just sell these things? Now, it doesn't seem like it's providing any actual at least on a sequential basis, right? And maybe things will kick back in the second half of the year, but I guess, just why not sell these IP addresses? I know the prices are down, but it's not like you're seeing good lease revenue growth.

and it obviously could help with the debt leverage. And listen, we are very focused on reducing our leverage. We understand that as a result of the Sprint acquisition, our leverage increased and there are really three major tools that we have to improve that. that leverage. The sale of assets is one of those. The growth in more profitable business is more impactful. And then finally, the reduction in costs. We look at the IP addresses and we have leased out approximately 15.2 million of the total 37.8 million that we have on a year over year basis.

The revenues associated with IP address leasing grew 18.1%. That's a pretty healthy growth rate. Yes, on a sequential basis it only grew at a half a percent. we will focus on growing those revenues. I don't believe in today's market we will maximize value by selling them. I do believe we maximize value. Michael's question around, you know, the remaining ten data centers and generating meaningful proceeds to deliver. So I actually view our path to delivering as working.

We are basically 6.2 times levered today. And that is a material improvement. I think that rate of improvement will accelerate due to further sales and growth in our EBITDA. know, as these integration programs complete, definitely by year end, but are substantially complete now that flows through. And the fact that 82% of our incremental sales in the quarter were on that, it allowed us a nearly 1% improvement in the entire installed base. We are definitely not back to where Cogent was pre-sprint, which was 76% on net and enjoying 40% margins without a subsidy payment from T-Mobile. We are keenly aware of of the fact that those subsidy payments from T-Mobile will end in about two years or less than two years. And, you know, we need to be able to grow EBITDA.

And I think the IP address increase Incremental leasing from this point will be a tool in helping us do that.

I hope to see that. And just one last one, Dave, if you don't mind. With the EBITDA declining this quarter, like at a time when you're trying to do the refi, what should we expect in terms of what that rate would look like so we can kind of factor that into our cash burn analysis?.

So as I said, our EBITDA actually sequentially improved quarter over quarter, which is It always does in Q2 because of the reduction in expenses. Right. But we did have these extraordinary expenses that were unique to this quarter with a six-month- It declined year over year. I'm just questioning how that impacts the refi and what rate you might get. Yes, so we have taken a number of steps to improve our financeability and cost of capital. One of those was the entry into the supplemental indenture and the expansion of our security capacity. are going to be replacing the unsecured notes with secure. That typically lowers your cost of capital. Offsetting that is the fact that our current secured debt is trading at a discount.

Witness the fact that we bought it back for 90.2 cents on the dollar resulting in a 13.4 million dollar gain in the quarter uh i think that's a indicative cost of capital so the yield to worse on those notes is about 8.8 8.9 percent We are working with bankers to determine the optimal enhancements to our notes to potentially lower that cost of capital. You know, I think it's premature for me to announce a rate. Ultimately, the market will set that. point is, you know, since those 2032 notes were issued, treasury rates on the comparable benchmark are up nearly, you know, a full percentage point. So it is absolutely reasonable that we're going to be paying more for our capital but i'm not in a position today give you a rate other than to look at where the current secures trade.

Thanks Dave, appreciate it. Dave Becker, CFO Alphabet and Google Hey, thanks Walt. And your next question comes from the line of Nick Dildale with Muppet Nathanson. Please go ahead.

Hey morning, thanks for taking my questions. First, Dave, on the Salesforce, it looks like a lot of the headcount reductions you had in the quarter affected the Salesforce. Sales productivity went up, so I assume that was concentrated in lower performing sales people but i guess prospectively you know as you're looking to you know obviously grow your revenues um How do you think about the current size and composition of the Salesforce relative to what you need? Okay.

Yes, so roughly about 40% of the headcount reduction was salespeople. We have been very disciplined about managing out under performers, and we had some remaining former Sprint salespeople who were here previously. primarily to transition and maintain the relationship with those remaining Sprint customers. We took a much more disciplined approach to managing those individuals out. They were underperforming They had been given, I guess, a pass on our normal discipline mechanisms and turnover rules and we implemented those more evenly in the quarter. I think in terms of the size of the sales force. It is probably needs to be about where it is today. you know, like one of my board members always jokes, you know, Stalin quote, there are fewer Russians after World War Two, but better Russians. I'm not sure I'm ready to go there with my sales force.

But, you know, the idea that we need to maintain productivity, you know, a metric that we look at probably even more important than, you the productivity as our cost per dollar of revenue acquisition ever Ever since the Sprint acquisition, that number had gone up. IT IS MATERIALLY COMING DOWN WITH THIS EMPHASIS ON IT IS MATERIALLY COMING DOWN WITH THIS EMPHASIS ON IT IS MATERIALLY COMING DOWN WITH THIS EMPHASIS ON PRODUCTIVITY AND ON-NET SERVICES. You know, I'm not in a position to give you an exact headcount number, but I think it's probably, you know, in the 500 range feels about right. So our corporate market is clearly growing slower than it historically had and has never fully recovered from the pandemic. And I don't see it probably recovering at this point. I think it's time to stop talking about the pandemic. Whereas with the added ability to sell wavelengths almost exclusively through the net centric sales team, you know, less than 5% of wavelength sales have been the enterprise and about 10% to corporate.

It's not zero, but it is. a very small base it's mostly net centric customers that's where we need to allocate more resources so i do think you'll see a continued shift Most of the Salesforce turnover has been on the corporate side. Okay, okay, that's helpful, Collin, thank you.

One other question, think about waves. Seems like everybody across the connectivity space is talking about a step up in demand from NeoClouds that really exploded on the scene in the past couple years. So I guess, how would you characterize the demand that you're seeing from that vertical? And what are you doing to make sure that your sales Salesforce is in front of those customers, many of which are newer, to educate about your services and be able to capitalize on that opportunity.

Yes, so we actually have a focused NeoCloud effort. We've identified those accounts. We've allocated them to more experienced reps. We have had great success with all of the house household names and I think we will increase our percentage of their purchases as we demonstrate the ability to provision and the ability of the service to deliver high reliability after its provision. You know, I'm always reluctant to mention names that sometimes can offend customers. But the companies that are rumored to be going public, those that are public are already today cogent wave customers. I encourage investors to do channel checks and reach out to customers. I can't disclose your name. You know the customer can clearly give an opinion on coaching.

What I will say though is. Most of the AI spend is announced, but not yet deployed. So even though, you know, So probably a trillion dollars of capital has been deployed in AI infrastructure. Only a small percentage of that trillion is actually in production being used for LLM creation or infrastructure. You know, there's a expectation that over the next four years there'll be a total of $7 trillion invested. But it's a little bit like a jigsaw puzzle where all the pieces need to come together before it's complete. And for many of these new clouds, they are either waiting on GPUs.

They are waiting on data center capacity. They are waiting on power. There could be rare cases. I know of one specific case of a large data center in North Dakota that's relatively the new oximetry network where the neo cloud is desperate to get wavelengths because they have power, but unfortunately there's no fiber today constructed between where their data center is constructed and the nearest network, turns out to be cogent to get them back to major markets. So, you know, I would expect in that case that when and it's a lot more than a mile is probably going to take a lot longer than, you know, three or four months to permit it and construct. We are not going to do that. It will get done and that backlog of waves will then be available. But they've already told us how many they need. we beefed up our network to deliver that.

I mean, that's just an anecdotal example.

Okay, that's great. Thank you, Dave. Hey, thanks, Seth. And your next question comes from the line of Frank Lauren with Raymond James, please go ahead.

Hey, Dave, this is Rob. I'm for Frank. Hey, so, you know, obviously, you know, you were just talking about the way of business. You spoke to it a bit here, you know, earlier remarks, you know, what, what are some things that, you know, you think you can do in order to drive more sales there as it definitely appears the market is rising and then, you know,.

what do you need to do to improve the legacy business from here? Yes. Hey, thanks for both questions, Rob. So, you know, I think first of all, The best way to win any business, legacy or wave, is provide the best value in the market. Now, as I've commented before, value can mean price, it can mean location, it can mean speed to install, it can mean reliability. And your reputation is built order by order. In many ways, one of the most encouraging numbers in what we reported is the fact that 77 existing wavelengths in the quarter were upgraded to larger numbers. capacity that kind of demonstrates as someone who dipped their toe in the water with cogent now is feeling comfortable to upgrade and take more locations you know You know, we do have the broadest footprint. We have the ability to provision quickly.

And because of our network architecture, we have greater reliability. 90% of Cogent's routes are unique to Cogent. We do know that for the areas of Sprint where they did not have fiber and we have used leased fiber, the fiber often comes from our wavelength competitors and the reliability on those leased routes is far below the reliability quality we deliver on our own routes. So I think it's just proving out all of those components. But again, while we're extremely transparent and granular, I would encourage you to talk to the couple of major competitors that we have for wavelengths and really find out if the wave demand that they are talking about is actually the wavelength that they're talking producing revenue today as opposed to wave demand that's effectively in the funnel. I totally concur that that wave demand is there. It's the question of converting it quickly enough and because Cogent is a new entrant, it's very visible. You know, we've gone from not even being considered on a third party ranking service to now being in the second tier of providers.

Our expectation is in the next year or so we'll break into that top tier and eventually we'll be one of. the two or three major nationwide wave providers. For the Cogent Legacy Services, I think there's three answers to the question. For all that services, there is a proliferation of fiber, which is allowing us to serve those locations, but we're going to do it with profit discipline and it is not our primary focus, but there are more locations. CoGEN has never sold off net services on non fiber infrastructure. We did inherit some from Sprint and then rapidly decommissioned as much of that as our contracts allowed. For our on-net footprint for corporate users, I will fully acknowledge that while the market has improved from the depth of vacancies at the pandemic, the office market across North America is still far weaker than ever. than it was pre-pandemic and number of work days in the office remain at about 60% of pre-pandemic levels. So, you know, while I think we are growing market share in our on-net footprint, we understand that that footprint has these structural constraints.

And then to that final point on that market segment, we don't intend to build into smaller or less traffic rich locations. And then finally, on our Netcentric business, which is by far and away the biggest part of our legacy business. it's about 46% of our total revenues and it's 98% of our traffic. There we continue to gain market share with 1,953 on net data centers. That, I think, is a testament to the breadth of our network. 308 markets, 58 And the fact that traffic grew sequentially 3% and year over year traffic accelerated to 16% is a demonstration of we're gaining share in the IP transit market and expect that to continue to grow and considering we're already already the largest player in that market. I think it's impressive that we're gaining share. If you compare our traffic growth numbers to either Cisco visual indexing or open vault data, reliable third parties, we're growing substantially faster, almost double the rate of the market. we're the biggest player in the market. I think that's still a testament to our ability to have market share again.

Great. Thanks, Dave. Hey, thanks, Raul. And your next question comes from the line of Anna Goschko with Bank of America. Please go ahead.

Hi, thanks Dave. I know this call is going long, so I'll try to just have some quick ones for you. So first of all, the proceeds from the data center sales on a net basis, I think was 224 million. And it looks like you spent 125 million to buy back the bonds at a discount, which is a good thing. But it leaves about 100 million simple math. So on that remaining $100 million of proceeds, are you going to go and try to buy back more of the existing secured at a discount? Or can you take that money and just repay the $27 million that's going to reduce the total amount of refi that you need to do? That's my first question.

Yes, and first of all, in many ways, Ana, you can ask as many as you want, because your question may be the most important to our investors on this call. So I think we are expecting to try to raise less than the $750 and use some of the proceeds to do that. We may be in the market to buy back some of the 27s while their discount is not as pronounced as it is on the 32s. I believe we have met our contractual obligation under the supplemental indenture to the holders of the 32s. We may also buy some additional 32s, but I think our primary objective over the next couple of weeks is to try to shrink the size of the new offering. Because to Walt's question, we understand it's going to cost us more and, you know, the kind of time value of the incremental payment versus where we capture an immediate benefit kind of washes out. So there's not really, I think, a big cash savings doing one over the other. think secondly, you know, to Michael's question earlier, we're also going to think about these additional proceeds.

And I think we've demonstrated to the debt markets, we are extremely committed to de-levering and we intend to use.

more proceeds to reduce debt. Okay, so that was my second, thank you. So that, was my second question. So the data centers that are still being marketed, Those are technically outside the restricted group for the dead. So I guess one, are you committing to use those sales proceeds to repay debt in the future? you addressed or mentioned that you're in discussion with potential underwriters about enhancements to what the new bond, So is that part of the enhancements that you're discussing? Is the commitment of future data center sale proceeds to the data group?.

Yes, I'll actually kind of answer those together, even though they were two questions. So you are absolutely correct that the data centers and the burn associated with them sit outside of the borrower group. So our intention when we. created this structure at the acquisition of Sprint was not to take collateral away from the current bondholders at the time, but rather to shield them from the carry cost of those datasets and be forced to use restricted payments capacity out of the borrower group to fund that burn. I think that strategy was bondholder friendly, and for the more sophisticated bondholders, I think they understand that. Then secondly, our willingness to go ahead and voluntarily contribute the proceeds into the borrower group. And remember, we contributed 100% of the net proceeds into the borrower group. While they're not all guaranteed to buy back debt, It definitely reduces net leverage at the borrowing group, and we've already spent, publicly disclosed, more than half, and it is our intention to use most, if not all, of that to just reduce gross debt as well as net at the borrower group.

In terms of the other data centers, we are definitely receptive to contributing those proceeds into the borrower group. To be candid, we have ample RP capacity to cover the burn. And since we have dramatically reduced our dividend expense, we have a great deal of flexibility. And that is something that ultimately will be part of the discussion between the underwriters. And, you know, while the underwriters are intermediaries, ultimately it's the whole team And if new bondholders felt that it was additive to make that affirmative commitment, it is absolutely something that the company will consider.

Okay, and then just finally, I think a bigger picture question for the whole structure. of more elevated this quarter. I think you already asked and addressed that. But you're not generating positive free cash flow. So obviously anyone buying kind of a new credit or a new piece of debt is going to wanna see that there's positive free cash flow to be able to service the debt. So could you just give us kind of the plan for getting to kind of sustain positive free cash flow?.

Yes, so I think it comes down to four inputs. First, our ability to grow desirable revenues. to our ability to continue to expand margins. I believe it is almost unprecedented. I have not seen an example ever of a public company that had 12 consecutive quarters of revenue decline And sequentially in 11 of those 12 quarters grew its EBITDA on an absolute basis in face of those declines. Now the tools we had to use were mostly cost savings. We were helped in this effort by the transition and subsidy payments from T-Mobile, and we understand that in two years, those payments go away. And since we're not issuing a two-year debt instrument, we need to be able to show that there is a... a path to free cash flow without those payments.

Inclusive of those payments, we are Okay, on cash flow, but not including them. It is challenging. So we do need to continue to grow EBITDA. Third, to kind of Walt's question around sales and trying to raise money, we are absolutely committed to that again. That is temporary. It is helpful. We have already demonstrated that we are going to take our leverage down. But, you know, I don't think we can asset sell our way to perpetual cash flow growth. And then finally, we need to not only generate free cash, and CapEx reduction is part of that story.

You know, we've had both year over year and sequential reductions. You know, we are in a challenging environment around price increases for equipment and it's unprecedented, but I do think we have some levers to pull to continue to be very capital disciplined. And again, I think our capital intensity per dollar of incremental revenue remains probably the best in the telecom. and I know that's a very bold statement to make, but I think if you look at Cochran's ROIC, since it went public in 2005, we've had an ROIC that's been substantially above our incremental cost of capital for that entire year. the entire period. We understand that not including the subsidies from T-Mobile that looked bad after the acquisition. We have worked diligently to spend that capital. You know, we took heat, for example, in spending $100 million on the 20, well, actually 125 Sprint facilities of which 24 were earmarked for sale. You know, we obviously got way more than that back by selling just 10 of that.

So, you know, we're going to continue to be very capital focused. And I do think that will. help us lower our cash burn and become cash flow positive more quickly. I know it was a long-winded answer to your question, but I get it. Free cash flow is what investors care about. Okay, great. Thank you, David. Hey, thanks, Anna.

And your last question comes from the line of Michael Rowlands with Citi. Please go ahead.

Thanks. Good morning, Dave. I'm curious to go back to your comments about the network traffic growth because I know acceleration to 16%. I'm curious if you could talk more about the significance of that between what you're seeing coming out of the corporate building portfolio or the multi-tenant buildings relative to data centers and what that means for the future, Q times Q math, to try to grow that transit revenue for Cogent on the annual basis. Yes, so two parts to the answer, Mike.

corporate business in the multi-tenant office on that buildings and off net and The demand enterprise business is irrelevant to our total traffic. It's only a couple percent of Cogent's traffic. 97, 98% of our traffic comes from data centers, from net centric customers. You know, our end user footprint is just not big enough relative to the size of our transit footprint. Now, within our transit customer base, we have about 7,500 access networks. We have about 6,000 content-generating businesses that drive that traffic growth and that 16% percent year over year. What we are seeing are two key things happening. Total traffic growth accelerating. But actually on a bit transferred basis, it's accelerating even faster.

So we bill our services on a peak utilization, either 90 or 95th percentile. So we throw out either 36 or 72 hours of peak traffic and bill on that variable. And we only bill on the high of the directions meaning in or out. For our access network customers, it's mostly in for our content producers it's mostly out. three quarters of all Cochran's traffic remains completely on our network. And if traffic patterns were not shifting, then the kind of difference between average and peak would not be relevant. But what is happening with the deployment of a genetic AI is we are seeing a market shift and end user traffic becoming materially more symmetric the result of that is for our access networks they're generating more bits that they are sending us, where previously they had sent us very low volumes. It may not immediately result in an increase in revenue, but over time it will.

And then conversely, for the content providers who are providing inference, they are now receiving. a lot of bets that they historically had not received, and those also do not immediately generate more revenue. But I do believe that as the internet reverse back to a more symmetric network, which was what it was at its inception, we we will see a meaningful correlation between traffic growth and revenue growth. Finally, the rate of price decline is moderating, and it is moderating in part because of the equipment constraints of the entire industry. And we are very fortunate in that we are only about 27% utilized in our IP network and have substantial inventory to sell without incremental capital where many of our competitors don't have that volume of inventory, have to go out and deploy capital, and in many cases, they can't even get the equipment if they want to deploy the capital. So I think it's all a good setup for us increasing our market share in that market.

Thank you. Okay. Thanks, Mike. And that concludes our question and answer session. I would like to hand it back to Mr. Dave Schaefer for closing remarks.

I want to thank everyone. I know today's call went a bit long, but I think these were extremely important topics for us to cover. I really appreciate investors' time and attention, and we'll be available at a few conferences to continue these discussions over the next week or so.

Take care all. We'll talk soon. Bye-bye. Thank you. And this concludes today's conference call. Thank you all for joining me now.

This live transcript is auto-generated without human intervention or review.

[Call has ended.]

Hi, everyone. My name is Sebastiano Petti and I follow the communications sector here at JPMorgan. It is my pleasure to introduce Dave Schaeffer, Founder and CEO of Cogent Communications. Dave, thanks for joining us today.

Sebastiano, thanks for hosting me. I'd like to always thank JPMorgan for a great venue. And importantly, thank investors for taking time out of their busy day.

Well, so just to start us off here, looking back, you closed the Sprint fiber acquisition in 2023. You laid out a multiyear vision, which was to convert data centers, launch waves, migrate more traffic on-net and ultimately emerge as a structurally different company. Now that we're 3 years in, I mean, what has played out as expected? And maybe what has surprised you along the path?

Yes. So I think there are 3 different parts to that answer. The first one is the conversion of the Sprint network into a wave-enabled network has pretty much played out as we expected in terms of the conversion. We initially set out to wave-enable 800 data centers. There have been additional data centers brought online and today, we're in 1,107 data centers.

I think the aggregate demand for those services turned out to be better than we expected and that there is a fourth incremental use case. That incremental use case is AI training. What has been disappointing has been the rate at which customers have actually accepted those waves.

Two, with regard to the data center footprint, our initial plan was not to spend capital on the 482 buildings that we had acquired with a gross inbound power capacity of 230 megawatts. About 1.5 years into the acquisition, it became clear to us that the power that was in place was a scarce commodity. We announced a plan to convert 125 of those 482 buildings into data centers.

Today, Cogent has a total data center footprint of 185 facilities, a total of 2.1 million square feet and about 213 megawatts of power. In that footprint, we identified 24 of the largest facilities for divestiture. We felt that we would not have the best business model to fill them up. We put those assets on the market, 10 of those facilities are under LOI to be sold, and we have another 14 that are earmarked for sale.

Third, in terms of the acquired customer base from Sprint, we anticipated needing to groom unprofitable services, noncore services, locations that were served with some technology, other than fiber. I think the rate of decline in that business has been faster than we had initially expected, but our ability to improve margins and take costs out have also been better than expected.

Cogent today operates on a global footprint. Its EBITDA margins absent the subsidies from T-Mobile or about half of what they were prior to the acquisition, but we've been able to demonstrate substantial margin expansion over the 2.5 years since the acquisition at an annual rate of about 800 basis points a year. We have said that our long-term margin expansion will continue but at a more moderate rate of about 200 basis points a year and should continue from the roughly 21% margins today until we resume margins above 40%.

Well, so great overview and definitely going to come back to a bunch of those topics. But just thinking about the last point there, just the guidance and our expectations over a multiyear basis, and the pullback in shares suggests some concern about the path forward. But again, you have, right, reiterated or reaffirmed your longer-term targets, not only on margin expansion, but also waves, as well as the deleveraging opportunity. Because what gives you the confidence in the long-term fundamentals of the business? And maybe what is the market missing?

Yes. So you are correct, Sebastiano. Our stock has not performed well over the past couple of years since the acquisition, and I think there is a concern among some investors around our aggregate leverage and our ability to refinance our 2027 debt. And I think we'll touch on that a little later in our discussion.

But in terms of the aggregate demand for the services that Cogent sells, the backdrop has probably never been better. Our core products are Internet-based services. 84% of our revenues come from those services. We are the largest carrier of Internet traffic globally. We operate in over 1,920 data centers in 57 countries, 306 discrete markets.

In that footprint, we have seen traffic growth reaccelerate to about 14% year-over-year from 10% year-over-year the last quarter. That acceleration is the beginning of AI inference and a new application driving growth off of a larger base. If you look at the Internet, over its 35-year history, it's experienced 5 previous episodic waves of growth. This is now the sixth wave of growth from a new application.

Secondly, in our wavelength business, which is a brand-new business for Cogent, we have gone from 0% to 3% market share in about 1.5 years. Now that is still far from the 25% market share that we anticipate ultimately capturing. But our footprint is the most ubiquitous, our routes are unique our ability to provision is quicker, our reliability is higher, and our cost basis in our network is lower, giving us a competitive advantage.

So what really gives me confidence in Cogent's long-term prognosis is our ability to expand margins due to on-net sales. Prior to the acquisition, Cogent was 76% on-net, 24% off. $1 of incremental on-net revenue carries 90% incremental margins. $1 of off-net revenues carry about $0.45 of incremental margin. And since the acquisition is closed, we have seen our product mix shift from 47% on-net back up to 62%. And in the most recent quarter, 83% of our sales were on-net.

So I think the combination of top line revenue growth coupled with margin expansion and our moderate capital intensity when compared to other service providers, the competitive footprint that we have and our more efficient infrastructure gives us confidence that we'll return to a pattern of returning meaningful capital to equity. We have returned almost $2 billion to our equity holders. We have paused that substantially due to our need to delever. And I think that has compounded some of the volatility in our share price.

Got it. Then I want to come back to the on-net in a minute here. But you alluded to it and just kind of going back to the data center sale, help us maybe think about the time line. So on the first quarter call, you spoke about the nonbinding LOI for the 10 data centers within potential or expected close in early summer. So again, walk us through the time line, when should we expect a binding agreement when maybe 8-K details on the counterparty and proceeds big area of focus for folks out there.

Yes. So we initially elected to convert the data centers in June of '24. By July '24, we had our first LOI. And then earlier this year, that LOI fell apart based on the counterparty requiring financing. We had another counterparty at the table that stepped in for a larger percentage of the portfolio. This is a well-capitalized 22-year-old infrastructure fund that manages $35 billion globally. They have completed their due diligence. We have negotiated a purchase and sale agreement.

Once that agreement is fully executed, you will see an 8-K from Cogent that will outline who the counterparty is, which 10 data centers, the exact purchase price and the anticipated closing date. We are still quite confident that we will close this in early summer.

Got it. And remind us of the intended use of proceeds from there, especially regarding 2032 notes and the planned issuance of new secured notes.

Yes. So just to remind investors, the public company, Cogent has no debt at the holding company level. Underneath of that, since 2010, we added high-yield debt at the operating entity, Cogent Group. And we operated up until the Sprint acquisition with that corporate structure.

With the addition of the Sprint network, we created a sister subsidiary Cogent infrastructure that holds the physical assets acquired from Sprint and also funds the associated burn with that asset. The data centers that are being sold exist at Cogent infrastructure outside of the borrower group.

We have committed to bondholders to take the proceeds of these first 10 data centers and entirely contribute them into the borrowing group, Cogent Group, and not increase our restricted payments capacity. We will then use a significant portion of those proceeds to repurchase some of our current secured debt provided that secured debt continues to trade below par.

After we have done that, we would then enter the market and issue new secured debt in order to replace the 2027 unsecured debt. In order to facilitate that, we are gathering a consent from the vast majority, more than 65% of our current secured bonds are held by companies that have agreed to allow us to increase our secured capacity, allowing us to refinance the entire unsecured tranche with secured capital, which should lower our cost of capital.

And are those the same as you think about the seeking consent from the secured bondholders to refinance the 2027 notes with secured paper. Are those the same bondholders that are likely to participate in the new deal? I mean, I guess what -- maybe any color on that?

I can't speak to each fund, but what I can say is the group of holders that represent more than the majority of our secured debt also hold a majority of our unsecured debt which will be called when there is no make-whole after June 15. They have indicated to us verbally that they would likely want to participate in that new offering. But we will work with a bank, potentially JPMorgan or some other bank in order to go to a broad marketing campaign and determine whether or not to allocate the new offering to the current holders or new holders.

Okay. And as we think about cash proceeds coming in from the data centers from this LOI, are there any material tax considerations or other deductions we should kind of be aware of? Just trying to think about what the proceeds could look like and your expectations around cash that you received?

You know what, we expect our cash tax leakage to be fairly minimal. While Cogent has approximately USD 1.1 billion in non-U.S. NOLs, predominantly in Europe, our U.S. NOLs are much more limited at about $140 million. We also will be spending capital this year and will be beneficiary of bonus depreciation accounting for that.

The combination of our NOLs and our capital expenditures this year, coupled with our current interest load and our ability to avoid any 163J limitations, I believe, will allow us to receive the vast majority, if not all of these proceeds tax-free, except for some local jurisdictions, which may not follow federal guidelines.

Understood. And after the data center sale and the debt repayment, talking about using the vast majority or a substantial amount of the proceeds, do you have an anticipated leverage ratio in mind? Or where do you expect things to kind of shake up?

So we have said that the company will continue to focus on delevering until its net leverage falls to 4x. We are today at 6.7x. Without disclosing the exact proceeds of the data center sale, this will be a material step in closing that gap between the 6.7 and the 4.0. We also will expect to continue to delever with the growth in underlying EBITDA.

If you look at the approximately 10 quarters since we closed the transaction with T-Mobile to acquire Sprint, our underlying EBITDA has grown at a little bit over $5 million a quarter sequentially. Now while that is not a perfect straight line, some quarters can be up $10 million next quarter could be up $1 million. The average is slightly over $5 million.

We anticipate our EBITDA to continue to grow at a similar or even better rate as the mix of on-net versus off-net traffic improves and as we return to total top line growth due to the complete attrition of the Sprint revenues that are negative or low margin.

Got it. And you read it. So as we think about reaching the 4.0 leverage target, in the past, you talked about before reaccelerating or revisiting capital returns. How do you think about -- or how are you -- what are the factors, I guess, that determine the pace and form of future returns, buybacks versus -- sorry, Dave, versus dividend?

So Cogent actually began returning capital to equity in 2007. It continued that capital return program entirely through share repurchases til 2010. At that point, we actually added high yield to our balance sheet, we had no debt previous to that, other than our capital lease obligations.

As we had grown our EBITDA, we implemented a return of capital or dividend program. We also continue to supplement that with buybacks. We had a dividend and we grew that dividend for 52 sequential consecutive quarters. With the capital that was necessary to repurpose the Sprint network and the decline in our top line growth rate, we decided to reduce that dividend to a minimal level of $0.02 per quarter or $0.08 a share a year. That dividend has mostly been counted as a return of capital. So therefore, has been tax efficient for the recipient.

We also have continued to do some episodic buybacks. In aggregate, we bought 10.3 million shares back at about $23 a share. In hindsight, we should have waited because the shares have traded below that, but no one can be perfect in their timing. We also have returned approximately $1.7 billion through the dividend, of which the vast majority was treated as return of capital.

As we approach 4x net leverage, we will commit to continuing to return capital to shareholders and whether the mechanism is buybacks or dividends, it will be highly dependent on market price of Cogent securities at that time.

Understood. Understood. So going back to just top of the funnel with the on-net commentary you touched on earlier. So with on-net plus waves was up, I believe, 9% in the first quarter and have been improving each of the last, I think, 3 quarters. As we're thinking about the drivers of growth, you talked about AI and some of these other things, maybe you can double-click on that for us and how much of this improvement is organic relative to the migration of the Sprint network, the Sprint off-net traffic on that?

So the migration of Sprint traffic from off-net to on-net occurred within the first several quarters of the acquisition and was a key driver in our ability to expand margins very quickly. When we acquired Sprint, it was 93% off-net and only 7% on. The Sprint business represented 42% of of the revenues of the combined company.

In that transition, we did improve margins, but subsequently, the margin improvement has come from continued cost reductions and synergies, but also from our ability to sell a much greater percentage of on-net than off-net. And on a going-forward basis, we anticipate more than 100% of our growth to come from Cogent's organic sales.

Now how can it be more than 100% because the Sprint revenue, which represents today, 16% of the combined company's revenue, down from 42% is continuing to decline. As it becomes a smaller and smaller part of our business, it becomes easier for us to have total top line growth.

Finally, the Sprint revenues have contributed some margin to Cogent, but their margin contribution is far below the average of the Cogent margin contribution. So without the payment subsidy from T-Mobile, our margins are around 21%. When we add that subsidy in, we're at around 31%. That subsidy will continue for just under 2 more years or about $200 million more in payments from T-Mobile.

As those subsidies wane, we expect to be able to continue to grow our underlying margins. And while our margins will expand meaningfully and our aggregate reported EBITDA will expand there will be in '28, a flattening of our rate of margin -- or rate of EBITDA growth due to those subsidy payments lapping.

Okay. And so the combination -- so I guess what gives you the confidence in achieving that? This year, I believe you talked about being able to expand margins over 200 basis points, which is -- 200 is the longer multiyear kind of target, but within 2026, it's -- do you think you can do better than that? Is this a function of just this accelerating on-net mix shift? Is there any -- help us think about any other kind of cost takeout opportunities that are maybe near term that we should be focused on as well?

Yes. I think there are actually 4 drivers of that ability to grow margins. The first and most significant is the growth in on-net versus off and that relative improvement in margins. Two, our ability to take the remaining Sprint business and price it appropriately. So each product has an acceptable level of margin. Third, we have still a small amount of synergies to be achieved from the acquisition. And those synergies will actually expand as we are successful in divesting of the unproductive data centers that I had mentioned earlier.

And then finally, we are still spending capital on integration work. Many companies stop when they get the easy parts of integration done, and then they wait literally years before they're able to show the savings. We have, I think, had the discipline to continue to focus on the hard parts of integration, and that's allowed us to continue historically raising our margins after that first year.

And then thinking about going back to just revenue real quick. In the first quarter, I believe revenue declined, you called out, I think, some large enterprise customers had canceled the month-to-month services. Acknowledging, I guess, the unpredictability of month-to-month contracts, should we anticipate sequential revenue growth from here? I mean that was something we've been talking about the last several quarters. So as you think about the on-net versus off-net mix that we just kind of described, do you anticipate continued sequential revenue growth from here?

So the rate of revenue decline has moderated every quarter sequentially for the past 5 quarters, each quarter being less than the quarter before. Secondly, that revenue decline was caused entirely by the Sprint customer base. And the organic Cogent revenues actually grew 28% in the 10 quarters since the acquisition.

It is really a question of will the Cogent revenue growth dwarf the rate of decline and the Sprint revenues. The Sprint revenues are today a much more de minimis portion of our total base at only 16%. But we are not giving a specific target, but I would anticipate the rate of decline to continue to moderate and may turn positive over this quarter or the next several quarters.

In the aggregate?

In aggregate.

Okay. Understood. That's great. And then moving to waves. So Waves installs slowed in the first quarter to the slowest -- to the lowest pace, I think, since the second quarter of 2025. You touched on customer acceptance issues, but also I believe there were some supply chain constraints as well. Maybe help elaborate on those and help us think about your anticipated -- when do conversion rates within the funnel, which we don't talk about anymore, when should conversion rates begin to improve?

So let me touch on supply chain issues 2 ways. One, their impact on Cogent; two their impact on customers. Cogent has built out a wave-enabled network over the entire Sprint footprint. We today have nearly 30,000 route miles of intercity wave network connecting over 110 markets. We also have over 21,000 route miles of metro fiber in those markets allowing us to connect to 1,107 carrier-neutral data centers.

The line systems, ROADMs and transponder shelves to deliver waves across that footprint are fully installed. We have an adequate supply of pluggable optics to add incremental wavelengths to that network. And what makes our network unique is the flexibility of being able to go from any data center to any data center.

So across that footprint, there's actually a combination of 10 to the 2,800 power of permutations of possible wavelengths ordered. We have sold wavelengths to date to 492 unique customers. We have sold those wavelengths into 581 of the 1,107 facilities. We actually installed more wavelengths in the quarter than we recognized revenue on. And many of our customers have had a number of constraints impacting their ability to accept those waves. Those constraints can range from limitations on power and the data centers in which they operate, their access to server equipment due to memory shortages and their access to pluggable optics.

So whether it be our IP-based services or our wavelengths customers still need something to plug those services into. And we have seen a number of supply constraints impact all of our customers.

Understood. And I think in an earlier session today, Verizon talked about AI infrastructure and -- something a topic that also came up last week, but talked about AI infrastructure as an opportunity for them. This is consistent with some of the channel checks we've done as well. But have you seen -- is there increased interest from the likes of AT&T, Verizon and others as well that's causing some pressure, whether it be on demand or pricing within the waves market?

So I would actually say the majority of the incremental use has come from hyperscalers. If you looked at the wave market historically for the past 15 years, there have been 3 major customer segments: regional networks looking to connect their networks together, international networks looking to extend their network and content distribution of information usually by hyperscalers.

The fourth use case, which has been the incremental driver has been the need to move data from one data center where data is stored to another data center where power is available for AI training. That has created a significant incremental demand on wavelength traffic.

For Internet transit, we have actually seen transfer traffic accelerate in part because the data that is collected over the Internet now has incremental value for training, whereas before it was discarded. So if we look at the 35-year history of the Internet, only about 20% of data transmitted was over stored, 80% was discarded. Now that ratio is inverted and over 80% of data transmitted is now being stored and used for training.

And we are now approaching the benefits of the AI training with inference. And inference means using those large language models that were developed in these large facilities, distributing them closer to the edge at the perimeter of the network. And then in an Agentic AI environment, 2 things are true. Total Internet usage by end users increases; and two, the directionality of that traffic changes to be much more symmetric as opposed to the asymmetry we've seen over the past 15 years.

Got it. And then just a quick follow-up on that. Just you modestly walked back maybe the mid-2028 time line on waves and reaching the 25% market share. But you still see that, I think, as the longer-term kind of target. But maybe help us why is 25% still the right number? And I guess despite the slower ramp perhaps.

So we operate in a fairly concentrated market. For metropolitan waves, the market is dominated by AT&T and Verizon. For intercity waves, the market is dominated by Lumen and Zayo. Cogent has 5 discrete competitive advantages, more endpoints, faster install, unique routes higher reliability on a per route mile basis and lower prices. We believe those competitive differentiations will allow us to replicate the market share that we have in global transit.

We are the largest carrier of transit globally, carrying just under 2 exabytes a day of information. representing about 25% of all global traffic.

Well, Dave, I think that's a great place to end it. Thank you for your time today, and thanks, everyone, for joining us.

Thank you, Sebastiano.

Good morning, and welcome to the Cogent Communications Holdings First Quarter 2026 Earnings Conference Call. As a reminder, this conference is being recorded, and it will be available for replay at www.cogentco.com. A transcript of this conference call will be posted on Cogent's website when it becomes available. Cogent's summary of financial and operational results attached to its press release can be downloaded from the Cogent website.

I would now like to turn the call over to Mr. Dave Schaeffer, Chairman and Chief Executive Officer of Cogent Communications Holdings.

Thank you, and good morning to all. Welcome to our first quarter 2026 earnings call. I'm Dave Schaeffer, Cogent's CEO. Joined with me on today's call is Tad Weed, our Chief Financial Officer.

A few key events and other significant matters in the quarter. I want to recognize some of the key achievements that we have made in the quarter. We have stated in previous calls, we intend to monetize 24 of our Sprint data centers that we acquired either via outright sale or leasing the acquired space on a wholesale basis. We have entered into a nonbinding LOI for the sale of 10 of these data centers. The counterparty has essentially completed its due diligence. Based on the status of this transaction, we expect closing to be early this summer. We continue to have multiple parties interested in other former Sprint data centers.

Now while we are working on refinancing our 2027 $750 million unsecured notes, which become due in June of 2027. At this time, we can make the following statement regarding the refinancing of our 2027 notes, and I'm going to ask Tad to read this statement.

Thank you, Dave, and good morning to everyone. The statement is as follows. The company and a limited number of holders of our 2032 $600 million secured notes who collectively hold more than a majority of the outstanding principal amount of our 2032 notes have reached a verbal agreement on a consent to amend the indenture for 2032 notes and that process is underway. If and once finally documented, the amendment will increase our ability under the indenture to incur pari-passu or junior lien secured debt and include several credit enhancements for our 2032 notes.

If and when the consent to the amendment is final, we will file an 8-K announcing the same and forgo our previously announced secured debt realignment plan. Please note that this discussion does not constitute an offer to sell or a solicitation of an offer to buy any security nor is it a solicitation of consent from any holders of our 2032 notes.

Back to you, Dave.

Thanks, Tad. We intend our refinancing to be complete after the expiration of our make-whole period which ends June 15, 2026. Once -- and if this transaction closes, our debt maturities will be as followed. Our current $600 million secured notes will mature in June of 2032. Our anticipated $750 million of secured notes will mature in [ 2033 ], $206 million of our secured ABS IPv4 notes mature in May of 2029. $174.4 million of our secured IPv4 notes mature in April of 2030. Whereas $629 million of IRU finance leases or capital leases have various maturities extending through 2046.

A couple of comments on our wavelength sales. At quarter end, we're offering wavelength services in 1,107 locations at either 10 gig, 100 gig or 400 gig capability. Our provisioning interval is approximately 30 days and continues to improve. Our wavelength revenues for the quarter were $13.6 million, an increase of 90.8% on a year-over-year basis and a sequential improvement of 12.3%. Our wavelength customer connections increased year-over-year by 71.2% and increased sequentially by 9.6% to 2,263. As of the end of the quarter, we have sold Wavelength services in 581 unique locations, and we have sold those services to a total of 492 unique customers. We intend to continue to focus on capturing 25% of the North American long-haul market. As of today, we have captured approximately 3% of that [indiscernible].

Now our EBITDA on a year-over-year basis, our EBITDA is adjusted for the quarter increased by $1.4 million, and our EBITDA as adjusted margin for the quarter increased year-over-year by 150 basis points. Our EBITDA as adjusted for the quarter decreased sequentially by $6.6 million to $70.2 million and our EBITDA as adjusted margin for the quarter was 29.3%. Seasonally, our SG&A expenses increased in the first quarter as compared to the fourth quarter. These changes are caused by annual CPI increases in salary, impact of payroll taxes in the U.S. the timing of employee vacations, our annual audit fees and our sales meeting.

Our SG&A increased from the fourth quarter of 2025 to the first quarter of 2026, by $7.1 million or 11%. By comparison, our SG&A increased by $10.6 million or 19% on from the fourth quarter of 2024 to the first quarter of 2025. This seasonal pattern is normal for Cogent.

We have a refined capital allocation strategy that is focused on delevering. We have committed the proceeds of the sale of our initial data centers that were formerly Sprint facilities to Cogent Communications Group, our borrowing entity, which will accelerate delevering at that entity.

Our total gross debt is adjusted for amounts from T-Mobile for the last 12 months on an EBITDA as adjusted basis for 7.4x EBITDA. Our net debt ratio was 6.79x at quarter's end. Our IP Leasing revenues increased 4% to $18 million and increased by 25% on a year-over-year basis. Our average price per IP address was stable at $0.40. We have titled to approximately $37.8 million IPv4 addresses and have leased out approximately $15 million of these addresses as of today. At quarter's end, we are providing services in 1,744 carrier-neutral data centers and 185 Cogent data centers. This footprint of data centers represents approximately 17 gigawatts of installed power. The Cogent data centers have approximately 211 megawatts of installed power and approximately 1.2 million square feet of floor space.

While our revenue growth for Q1 2026 was negative, the decline in revenues from acquired Sprint customers is moderating. We anticipate a long-term average revenue growth rate of 6% to 8% and EBITDA margin expansion of approximately 200 basis points per year. Our revenue and EBITDA guidance targets are intended to be multiyear and are not intended to be quarterly or annual specific items.

Now I'd like to turn the call back to Tad to read our safe harbor language, provide some additional detail, and then I will provide some summary remarks and open the floor for questions and answers.

Thank you, Dave. This earnings conference call includes forward-looking statements. These forward-looking statements are based on our current intent, beliefs and expectations. These forward-looking statements and all other statements that may be made on this call that are not historical facts are subject to a number of risks and uncertainties, and actual results may differ materially. Please refer to our SEC filings for more information on the factors that could cause actual results to differ. Cogent undertakes no obligation to update or revise our forward-looking statements.

If we use non-GAAP financial measures during this call, you will find these reconciled to the corresponding GAAP measurement in our earnings release that are posted on our website at cogentco.com.

Summary of results. Comments on our revenue mix since the Sprint closing, which as reminder, was May 1, 2023, so the first full quarter with Sprint combined with Cogent was the third quarter of 2023. Despite revenue decreases, we have been able to increase our margins, our increases in gross margin and our EBITDA margin have been driven by cost reductions and a rotation to our more profitable on-net products.

Comparing our revenue by connection type from the third quarter of 2023, again, the first full quarter when we were combined with Sprint to this quarter illustrates the material change to the composition of our revenues and the strength of our underlying Cogent classic business. Our on-net revenues were 47% of our total revenues in the third quarter of 2023. Our total on-net revenues including wavelength on net revenues, increased from 47% to 62% of total revenues this quarter. Our less profitable off-net revenues were 48% of our total revenues in the third quarter of 2023, and our off-net revenues have decreased to 37% of our total revenues this quarter.

Lastly, our noncore revenues were 5% of our total revenues in the third quarter of '23, and our noncore revenues have decreased to $1 million and were approximately 0.5% of our total revenues this quarter. Our total revenue for the quarter was $239.2 million. Our total revenue for the quarter declined sequentially by $1.3 million or by 0.6%. The decrease was a slight improvement from the $1.4 million sequential quarterly decline last quarter. USF tax revenues had a negative impact on our sequential revenue results of $0.3 million and a negative impact year-over-year $0.7 million.

For the quarter and sequentially, our on-net revenues, including on-net wave revenues increased by $2.8 million. Our less profitable off-net revenues declined by $3.9 million. Our noncore revenues decreased by $12.2 million. Our wavelength revenues, which is entirely on-net, increased by 1.5%.

Our gross margin percentage for the quarter increased year-over-year by 150 basis points to 46.1% from continued cost reduction and product optimization, including our focus on on-net products.

Some comments on revenue by class. We analyze and classify our revenues into 4 network connection types and 3 customer types. Our 4 network connection types are on-net, off-net, wavelengths and noncore. Our 3 customer types or NetCentric, corporate and enterprise customers. The substantial changes in the acquired Sprint wireline revenue base have masked the underlying performance of our Cogent classic business. Our consolidated revenue declines have been largely attributed to the reduction in the acquired Sprint wireline corporate and enterprise noncore and off-net revenues.

At closing, the Sprint wireline revenues were 42% of our total revenues. That percentage has declined from 42% to only 16% of our total revenues this quarter. We acquired Sprint wireline with a revenue run rate of $118 million per quarter. This acquired revenue base has decreased from $118 million to $39 million this quarter run rate. That represents a $79 million reduction in quarterly revenue related to the acquired Sprint revenue base or a 67% decline since deal closing. At deal closing, our Cogent Classic revenue run rate was $155 million per quarter, and that run rate has increased by 28% from $155 million to almost $200 million, $198 million for this quarter.

Our total corporate business represented 42.3% of our revenues this quarter. Our quarterly corporate revenues decreased by 8.7% year-over-year and sequentially by 1.7%. The Sprint wireline corporate revenue customers represented 30% of our total corporate revenues at closing of the acquisition, and those Sprint acquired corporate customers now represent only 10% of our total corporate revenues. The Sprint Wireline acquired corporate customer base has decreased from a run rate of $39 million per quarter at closing to a current run rate of only $8 million for this quarter, an approximate 80% declining.

Our total NetCentric business continues to increase and to benefit from the benefit from the growth in video traffic, activity related to artificial intelligence, streaming, IPv4 leasing and wavelength sales. Our NetCentric business represented 44.2% of our revenues this quarter. Our quarterly NetCentric revenues increased by 14.2% year-over-year and sequentially by 2.3%. The Sprint wireline NetCentric customers represented 21% of our total NetCentric customer revenues at the closing of the acquisition.

Those Sprint Wireline acquired NetCentric customers now represent only 6% of our total NetCentric revenues. The Sprint wireline acquired NetCentric customer base has decreased from a run rate of $19 million per quarter at closing to a run rate this quarter of only $8 million and approximately 60% decline, 58% actually.

Enterprise business. Our total enterprise business represented 13.5% of our revenues this quarter. Our quarterly enterprise revenue decreased by 26% year-over-year and sequentially by 5.7%, primarily due to a reduction in the acquired Sprint Wireline enterprise off-net revenue. The Sprint wireline enterprise customers represented virtually all of our enterprise revenues of the closing of the acquisition and the Sprint Wireline acquired Enterprise revenue base has decreased from a run rate of $60 million per quarter at closing to a current run rate of $23 million, a 62% decline.

Revenue and customer connections by network type. We serve our on-net customers in 3,605 total on-net buildings. Our total on-net revenue, including on-net wavelength, was $149.2 million for the quarter. That's a year-over-year increase of 9.1% and a sequential increase of 1.9%. Our less profitable off-net revenues was $89 million for the quarter, a year-over-year decrease of 17% and a sequential decrease of 4.2%. Our off-net revenue results are impacted by the migration of certain off-net customers to on-net and the continued grooming and termination of low-margin off-net contracts, virtually all of the decline from the Sprint wireline acquired customers.

Our average price per megabit for our installed base decreased sequentially to $0.12 from $0.14 last quarter and was $0.20 for the first quarter of last year, and our average price per megabit of new contracts for the quarter was $0.07 compared to $0.06 last quarter, so a slight increase and $0.10 in the first quarter of last year.

Our ARPU for the quarter were as follows: our on-net IP ARPU was $514, our off-net IP ARPU was $1,219, our wavelength ARPU was $2,093, our IPv4 ARPU was $0.40 per address.

Churn rates. Our on-net churn rate was stable and our off-net churn rate actually slightly improved from last quarter. Our on-net unit churn monthly rate was 1.2%, the same as last quarter. Our off-net churn rate is primarily driven by the reduction in the acquired Sprint customer base, and it was 1.7%, moderation from 1.9% last quarter. Our wavelength monthly churn rate is less than 0.5%.

Traffic. Our year-over-year IP network traffic growth continued for the quarter. Our IP network traffic for the quarter increased sequentially by 4% and increased year-over-year at an accelerated rate to 14% this quarter compared to the same quarter last year.

Sales rep productivity. Our sales rep productivity was 4.1% this quarter, the same as last quarter and compared to our long-term average of 4.8%.

FX. Our revenue earned outside of the United States, about 21% of our revenues this quarter. Based on the average euro Canadian conversion USD rates, so far this quarter, we estimate that the FX conversion impact on sequential quarterly revenues will not be material and the impact on year-over-year would be a positive of approximately $1 million. Our revenues and customer base are not highly concentrated. Our top 25 customers represented 16% of our revenues in the quarter.

CapEx. Our capital expenditures were $46.2 million in this quarter. We have experienced multiple equipment price increases from vendors due to supply chain constraints so far this year. Our principal payments on capital leases were $13.4 million this quarter.

Debt and debt ratios. Our total gross debt at par, including $629 million of finance lease IRU obligations, was $2.4 billion at quarter end, and our net debt total net of our cash our $181.7 million due from T-Mobile was $2 billion. Our leverage ratio, as calculated under our more restrictive unsecured $750 million 2027 notes indentures that we plan on refinancing was 6.1% our secured leverage ratio was 3.79%. Our fixed coverage ratio was 2.29%.

The definition of consolidated cash flow under our $600 million secured 2032 notes indenture includes cash payments under our IP Transit Service agreement with T-Mobile in the determination of consolidated cash flow. Our anticipated $750 million secured notes indenture will include the same definition of consolidated cash flow, again, including cash payments under IP Transit agreement. Our leverage ratio, as calculated under our $600 million Secure 2032 notes indenture was 4.66%. Our secured leverage ratio was 2.9% and our fixed coverage ratio was 3%. Lastly, on bad debt and day sales. Our days sales was 31 days at quarter end. Our bad debt expense was less than 0.5% of our revenues for the quarter.

And with that, I will turn the call back over to Dave.

Okay. Thanks, Tad. I would like to highlight a couple of strengths of our network, our customer base and our sales force. We are direct beneficiaries of increased demand for over-the-top video, AI activity and streaming video trends. At quarter's end, we were able to sell wave services in 1,107 data centers across North America with a reduced provisioning interval of approximately 30 days. We are selling Wavelength services as of quarter end to 492 unique customers and 581 unique data center locations. At quarter-end, we are selling IP services globally in a total of 1,929 data centers. At quarter end, we are directly connected to 7,630 networks. 22 of these networks represent settlement-free peers, 7,608 of those networks are Cogent transit customers.

We remain very focused on our sales force productivity and managing out underperforming reps. Our sales force turnover rate was 4.8% per month for the quarter, below our historical average of 5.7% per month. At quarter end, we have a total quota-bearing sales force of 568 reps. This includes 285 professionals focused on the NetCentric market, 269 sales professionals focused on the corporate market and 14 sales professionals focused on the enterprise market.

We've made significant progress in several areas. We're improving our margins and growing our EBITDA due to our diligence in cost reduction and our focus on selling more profitable on net services. In the first quarter of 2026, 83% of our sales were on-net services. This increased the percentage of our total base to 62% of all services being on-net. We have a clear path to refinance our 2027 $750 million unsecured notes with secured $750 million notes. We are actively working to continue to monetize former Sprint facilities, and we are looking to grow EBITDA, which will further accelerate our delevering and allow us to reaccelerate our return of capital program to equity.

We're optimistic about our wavelength services business. Our wavelength services are differentiated in quality of service, breadth of footprint, uniqueness of routes and efficiency and provisioning. Our on-net services, both IP and wavelength, offer unparalleled value to customers. We offer superior services for all of our products, a broad footprint in revenue rich locations, expedited provisioning and disruptive pricing. In summary, we win on value.

Now I'd like to open the floor for questions.

[Operator Instructions] Your first question comes from the line of Greg Williams from TD Cowen.

Great. Dave, the first one just on EBITDA. It was a touch light versus the Street in our estimates. You mentioned that obviously, you have the seasonal cost, payroll taxes, CPI, et cetera, and it's up $7.1 million quarter-over-quarter. How much of that $7.1 million was the seasonal cost? And maybe talk to the cost takeout progress. Essentially, I'm just trying to figure out the EBITDA cadence next quarter and the balance of the year. Are you still looking for 200 bps of expansion or I think you said greater than 200 bps this year?

And then the second question is just on that data center sales process. You mentioned the 10 data centers you're looking to close this summer. Any color would be helpful in terms of valuation price per megawatt. Is it coming close to the $10 million a megawatt? Maybe just generally characterize them versus the other 14? Are they better, same, worse, larger, smaller? Any help.

Thanks for both questions, Greg. So first of all, in terms of EBITDA margin expansion, we historically experienced a reduction in EBITDA margin and an increase in SG&A expenses in the first quarter. This pattern has been in place for 20 years as Cogent has been a public company. The increase this quarter was approximately $7.1 million. The vast majority of that increase will go away, and we expect to be able to resume our sequential increase in EBITDA margins as well as our year-over-year expansion. And while we will probably not repeat the roughly 800 basis points of improvement last year, meaning 2025 over 2024, we do expect to be over our multiyear guidance of 200 basis points on a year-over-year basis.

With regard to the 10 data centers, the aggregate proceeds are substantially more than the $144 million. These 10 represent a pretty good average across the 24 data centers that we are looking to divest of. It does not include our largest data center or our smallest data center that we are looking to sell. We also have a number of other parties conducting due diligence on multiple other data centers. We are focused on getting this transaction completed early summer. And then using those proceeds to be able to rapidly delever at the Cogent group level.

Just to remind investors, the data centers are held at Cogent Infrastructure, which is not a borrower under our high-yield indentures. We have committed and we'll continue to commit to contribute the proceeds of these 10 data centers that are being divested of to the borrower group and then use that money to rapidly delever both on a gross and net basis.

Your next question comes from the line of Sebastiano Petti from JPMorgan.

Dave, I think last quarter, we talked about hitting an inflection point where the growth in the organic business would offset the spread declines, but despite favorable currency, I guess, sequentially here, the business did contract. I mean, just help us think about any onetime anomalies in the business? How should we think about the top line trajectory from here? I think Tad talked about it being neutral on a constant currency basis as we think about the second quarter.

And then any update on the Wave installs just slowed a little bit sequentially here. Is this related at all to the supply constraints that Tad talked about in his prepared remarks?

And I guess relatedly or just to kind of confirm the 25% market share target in waves, is that still anticipated by May 2028? Or should we anticipate the timeline has been extended because it didn't seem you were specific in your prepared remarks.

Yes, sure. So the inflection in revenue was related almost exclusively to several large enterprise customers churning a portion of their off-net revenues. While those were not anticipated, those revenues were out of contract and on a month-to-month basis. The core Cogent business and the on-net business in totality grew both sequentially and year-over-year. Our primary focus is on growing on-net revenues, 83% of all revenues sold in the quarter were on-net, and that will help us increase our aggregate profitability and our free cash flow and EBITDA.

The wavelength install rate was not impacted by our supply constraints, but it was impacted by supply chain constraints of our customers. We have not, as of yet, began to force build wavelength services. I think this is part of the way we've been able to grow both the number of locations and number of customers that we sell to. The supply chain constraints did hit Cogent in terms of capital equipment from plugable optics to normal sequential capital installs across our network.

All of our major vendors have had price increases. Actually, our primary vendor had 4 price increases in the 4 -- in the first 4 months of the year. This is counter to a pattern of prices for technology declining.

On wavelength installs, we have seen a variety of customers pushing out their acceptance of wavelengths. We actually provisioned more wavelengths in the quarter than we did in the previous quarter, but the customers did not accept them. That decision to push out acceptance is being driven by constraints. We have seen constraints of power availability in data centers. We've seen customers actually change wavelength termination points to avoid a constraint in one data center in a market moving to another data center. There are equipment constraints from plugable optics on the customer side to the ability to have GPUs installed to accept wavelengths.

And probably something that should be obvious that people forget is while there is a rapid acceleration of capital for AI training and there have been literally hundreds of billions of dollars annually of announced investments and trillions of dollars in total, almost all of those announcements are not yet online. And in many cases, customers order wavelengths to facilities that are not yet either fully powered or fully constructed. We do think that will ease.

With regard to our ability to gain market share, we have gone from 0% market in 2 years to 3% of the market. Our goal remains to hit 25% of the intercity long-haul market. We feel that is very reasonable based both on the number of locations and the diversity of the customer base. While we are hopeful that we can reach that by mid '28, that is just a little over 2 years from now, and these equipment supply constraints may, in fact, impact that, we are not in a position to make that determination.

Your next question comes from the line of Chris Schoell from UBS.

You mentioned the equipment prices stepping up from vendors due to the supply chain constraints. But is there anything else causing CapEx to come in higher than that $25 million per quarter run rate you previously spoke to? And should we assume this level of capital spending will persist in the near term?

And then maybe just one on the sales force. It appears that head count has been stepping down consistently. What is the main driver there? And do you believe you can still hit your revenue targets with a lower headcount?

Yes. Thanks for both questions, Chris. So first of all, on equipment pricing, I think there have been 2 primary drivers that are forcing vendors to raise pricing. The first is the acute shortage of DRAM. And since DRAM is utilized both in optical transport and routing equipment that is causing our vendors to experience a higher cost of goods sold. The second has been a shift in buying patterns. So historically, service providers represented the vast majority of equipment purchases. Those equipment purchases have become concentrated in a handful of hyperscalers that have exerted very strong pressures on gross margins for our vendors. In order to offset that margin pressure, our vendors have increased prices on service providers while offering the aggressive discounts for volume to hyperscalers.

We don't have enough data to know how all or how material these trends will be going forward, we do expect our capital intensity to continue to moderate. However, these increases in pricing were not anticipated and are not in line with historical trends. This is the first time in Cogent's 26-year history that we've seen the prices of our key technologies increase, not decrease. We do think these are not permanent, but we don't have enough data to fully answer that with conviction.

With regard to headcount, we have tried to manage out unproductive reps. And we have consolidated some teams in order to better affect training. We do believe we will see an acceleration and rep productivity, while on a unit basis, it remained flat, on a dollar of revenue acquired basis, it actually improved materially both sequentially and year-over-year. We expect to see an improvement in rep productivity, both on a unit basis and dollar of revenue acquired. We also are continuing to hire reps and believe that the vast majority of the housekeeping that we have done is behind us now, and we should be at a point where the sales force will stabilize and then begin to resume growth as there is adequate addressable market for our services to allow us to support a larger number of sales reps across all of our products, but holding reps accountable to productivity targets is critical to our ability to hit our margin objectives.

Your next question comes from the line of Ana Goshko from Bank of America.

So Dave, a few questions, follow-ups. So just on the timing of the data center sales, so you're still at a letter of intent and you said that you expect to close the sale early summer. So that seems like a pretty fast turnaround. I think in the past, you had said it might even take like several quarters to close the deal from the actual agreement.

So a few things. When do you expect that the actual sale agreement will be finalized? And when that happens, will you press release or 8-K that for us with the dollar amount? And then, yes, I just want to confirm, when you say early summer, what does that really mean in terms of July or late June?

Yes. So first of all, the counterparty has been actively completing its diligence with a battery of consultants. They have spent several million dollars on that diligence and it all has been confirmatory. And they have indicated that they would like to accelerate the closing once we have a final purchase and sale agreement in place, they have agreed to shorten the period of time from their LOI expiring to closing. We expect that to be in early summer which would mean probably June or early July at the latest. We will announce the economics and the locations in an 8-K once the deal has been put under a binding agreement with a nonrefundable deposit, and we will disclose the name of the counterparty as well as the exact proceeds.

And finally, we have committed that those proceeds will be contributed to Cogent Group, the borrower and that those proceeds will be earmarked for net delevering. And in some cases, a portion of those proceeds will also be used for gross delevering.

Okay. So just to put that maybe in kind of simpler terms. So are you saying a portion of the proceeds will go to pay down debt, but not all of them?

So we have committed to a group of bondholders that the vast majority of the proceeds will go to buy back debt but we did not commit to a number that equals the purchase price. We just committed to a number that is a significant percentage of what the final purchase price would be since we did not disclose that information to the bondholders as it is nonpublic at this time.

Okay. And then just another follow-up on this. So you had said that you plan to refi the unsecured with the new secured the full $750 million after the call price drop in June 15. So have you thought about if these proceeds are going to be coming in so soon, why do you need to do the full $750 million? Could you do a smaller deal and then just use the proceeds to pay down a portion of the bonds that are due in '27, the unsecured?

So the answer is we can do that. But with our current $600 million secured debt trading at a discount to par, we want to try to capture some of that discontinuity and buy back the current $600 million secured until they trade closer to par. And then at that point, the additional capital that we have could be used to result in a smaller new issuance. But today, the current secured debt is trading at a material discount.

Okay. And then finally, just a quick follow-up on the business model. So on the cost side, you had previously talked about there being $10 million of annual synergies left from the Sprint acquisition, then also that there were integration costs of about $3 million a month that should be rolling off this year. So if I put that all together, it's roughly maybe like $45 million of annualized cost saves that you could theoretically or hopefully, in practice achieve this year. Just wanted an update on where you stand and what that outlook for the actual cost reductions looks like this year?

So we have achieved a small portion of that $10 million in remaining synergies. And just to remind you, we actually increased that target after we had already achieved the initial targets that we had laid out. The remaining integration work is continuing. That number was running as high as $5 million a month or $60 million a year. Today, it is slightly below $3 million a month, and we expect both of those areas of savings to be complete by year-end. We have not disclosed the exact pacing throughout 2026. But all of these roughly $45 million in costs will disappear in 2027.

Your next question comes from the line of Frank Louthan from Raymond James.

On the wavelength, what's the average size wave that you're selling now currently? And then on -- were there any dark fiber or IPv4 address sales that helped contribute to revenue or EBITDA this quarter?

Frank, thanks for the questions. I'm going to actually take those in reverse order. There were no dark fiber sales in the quarter, while there was some IPv4 unit activity. Actually, the number of IP addresses leased went down slightly sequentially but the revenue went up. It is a combination of selling at higher prices and continuing to raise prices on legacy orders. We do anticipate continued growth in our IPv4 business. We do not forecast any dark fiber sales. We treat those on an episodic basis and do that only to date with parties as a way to help them out of a bind if we have a route that is particularly critical to their operations. Many -- or to date, the handful of dark fiber sales that we've done have been 2 counterparties where we have been a customer of theirs buying dark fiber for a number of years.

With regard to Wave sales, the vast majority of our waves have been 100 gig waves. We are seeing an increase in the number of 400 gig waves and a significant decrease in the 10-gig wave. I think an average number would be somewhat misleading. I think looking at that on a model basis is the best way to do that. And roughly about 75% of our sales have been 100 gig waves.

Yes. That's kind of how I was thinking about it. So 75% of 100 gig waves, what was that last quarter? And of that 25%, are they substantially 400 gig wave? Is that the way to think about it? How much is that 400 gig wave growing as a percentage of your new sales?

Yes. So the 100 gig percentage has remained relatively constant. I think it was 78% the quarter before. And of the remaining 25%, there definitely has been a shift away from 10 gig sales and a shift towards 400 gig sales with today, over 10% of sales being 400 gig sales.

Your next question comes from the line of Walter Piecyk from LightShed Partners.

Dave, on these data center sales that are coming up, I guess just getting back to Ana's question. Is there anything that forces the buyer to act by a certain period of time? Otherwise, is there a risk that they understand the dynamics of your the refi coming up on the $750 million in June of next year and try and push that out as a leverage point to impact price. Can you just give us a little bit more on those terms as well as what is it that you had to consent pay off, whatever it is on the 2032 noteholders to get them? I think you were saying to effectively enable the refi in the '27, I don't necessarily understand that connection if you could put a little bit more color on that.

Yes. Sure, Walt. So first of all, we entered into a letter of intent that has exclusivity for the intended buyer on these data centers. That period of exclusivity ends, those -- some of those facilities have backup agreements that are ready to spring into force if the exclusivity period is allowed to lapse. So there is a significant amount of pressure on the buyer to inoculate themselves from a counter offer.

Secondly, they have spent in excess of $3 million on diligence, and they are in the process of rounding out their management team to absorb these facilities. So we believe they are going to move forward, but the biggest lever that we have is a counteroffer that would spring in if their exclusivity period lapse. They have indicated to us that they actually want to shorten the window from the expiration of that exclusivity period to closing. And that's what gave us confidence and our decision to announce early summer rather than later in the summer.

To pivot to your second question about the current 32 note bondholders, while we have every right under our indentures to do the IRU realignment that we disclosed on our last earnings call, in discussions with many of those bondholders they preferred a more traditional way of giving us the flexibility to increase our secured leverage. While we have a substantial amount of capacity for additional leverage, we were constrained by the 4x net leverage limitation that's embedded in the 32 notes, and it will be their decision to increase that to allow us to fully refinance the $750 million with a single unitary secured issue. And in doing that, they would then be in a position to see us not realign the IRUs and keep those with their associated debt and the borrower group.

And do you anticipate -- if all goes well, you sell the data centers ahead as you told Ana, take down secured, refi the $750 million, what type of rate do you anticipate? I think you're paying 7%on that now, same rate low or higher?

Yes. We are paying 7% on the current unsecured bonds. Our current secured bonds are trading at just around 8% today. I believe our new issue will most likely price off of the trading of those bonds and will be somewhat similar. We'll have both a new issue concession. That's typically about an 8-point and it could have a small variance based on duration. If we sell the data centers use a portion of the proceeds to buy back bonds, it is likely that the current secured bonds will trade asymptotically to par, which is 6.5% and then it would allow us to finance probably at a similar rate.

While I can't predict the exact trading off the bonds, the sequencing of completing the data center diligence period, converting into a binding agreement, announcing it along with the announcement that Tad mentioned around the exact mechanics of our agreement with the majority of the bondholders and then earmarking the exact amount of dollars that will go to repurchasing bonds of the current 32. And then to Ana's point, maybe a portion of that money maybe held in advance and just allow us to refinance a slightly smaller amount of money than the $750 million, that's outstanding.

But all of this is designed to drive down our cost of borrowing and make our new bonds similar to where our existing bonds are, which I think is an achievement considering the aggregate increased cost of capital since those bonds were issued.

I mean my guess is operational performance like sequential revenue growth and wavelengths growth will probably have a bigger impact on where the secured debt trades relative to some asset sales relative to a much larger debt load. But I guess what would be helpful is understanding why would unsecured trade at parity was secured?

Well, today, the unsecured actually trade at a discount to secure. Our current unsecured trade at roughly 7.1 and our secured trade at about 8.1.

Why do you think that is?

I think it's primarily duration.

Okay. But then the rate market will obviously have an impact. Just one last in terms of understanding cash burn. The CapEx, I think you said last year or 2026, not excluding capital lease obviously, should have been about $100 million for this year, a big cutdown for the variety of reasons that you guys have talked about. You were at $46 million for the first quarter. Is it just going to drop off a cliff in future quarters? Or should the CapEx run rate maybe be higher than the $100 million that you talked about?

Yes, Walt. So our CapEx on a Q1 '25 to Q1 '26 dropped by about $13 million. So it dropped from roughly $59 million to $46 million.

But up sequentially.

Well, typically, declines in fourth quarter and steps up in Q1, again, like our SG&A. We do anticipate our CapEx coming down on a year-over-year basis. But as I mentioned and Tad mentioned in the prepared remarks, we have been shocked by the fact that our equipment vendors have actually raised prices, which is highly unusual in a technology business. We think those may be over. And if they are, we'll be much closer to the $100 million number. If there are future increases in equipment that will push up our costs, primarily for plugable optics, which are probably the largest single item that we spend capital on.

Okay. Just one last follow-up, Dave. Just again, going back to the Ana's question. She's obviously as a data analyst a lot smarter about this stuff than I am. What -- I guess if it's trading at a discount today, right, and you're like, oh, if it's a par when it's time to refi, like why bother then with the secured note if it's at par, then take a smaller unsecured note out. I mean, shouldn't that..

That may be the case, Walt. That's why I said we will look to capture discontinuity while the current secures are trading at a discount.

Your next question comes from the line of Tim Horan from Oppenheimer.

I think, Dave, on the data center sale, the binding sale agreement, does that have to occur like a month before the final close? Or can you give us some color around that? And can you talk about a little bit more color where you are with selling the other data centers roughly will be the same price? And when do you think you'll be able to kind of have a letter of intent?

Yes. Thanks for the question, Tim. So first of all, the period between contract signing and closing has actually been shortened at the purchasers request. Normally, you would have a window of up to 90 days from binding agreement to sale. This is substantially shorter than that, and that's what gives us confidence that we will end up closing this in early summer. And then in terms of the other data centers, we are in discussions with multiple counterparties, some for just one facility, some for several, some are in kind of a backup position to the current party and we've informed them of the likelihood that the current party is moving forward. We have tried to focus our data center resources on getting this initial 10 centers over the finish line. And then for the remaining 14 we will hopefully be in a position to work more expeditiously to getting some of those deals move along, but we've really tried to keep resources focused on getting this deal closed.

And then, Dave, on the wavelength side, could you give us your best guess then on when you think you can hit 25% share? And can you just talk about the dynamics of where you are winning share? Is this -- is this new builds? Is it when contracts expire? Is it moves? Or is it because they're increasing from 10 meg to 100? Any more color would be great.

Yes. So first of all, it's kind of all of the above in terms of our wins. We are winning existing waves with customers that are frustrated with their current supplier. We are winning waves from customers who are increasing their throughput. We are winning waves due to locations shifting and the breadth of our footprint. And we are winning brand new builds, particularly from hyperscalers and [indiscernible] clouds which are new to the market.

With regard to getting to a 25% market share, we feel very confident that we will achieve that level. Doing it in a little over 2 years, does become harder as we see the current rate of installs not being accepted by customers. We are working as diligently as we can to install if customers are ready to accept. I think we'll need another quarter or 2 to be able to definitively answer that question. But we do see a significant pent-up demand for the locations, the routes and the price points that we are offering.

Your next question comes from the line of Nick Del Deo from MoffettNathanson.

Turning back to CapEx. Just to be clear, was all the increase this quarter versus what you've guided to attributable to higher prices? Or were there more units of equipment you purchased or inventory build, anything like that going on?

Yes. So as I mentioned in Walt's answer, our CapEx was down $13 million on a year-over-year basis. It's -- that was probably about half of the level of reduction that we would have anticipated and would have been kind of on plan. I would say that the majority of the overruns came from price increases, but there was also some preordering of equipment that we're concerned about delivery schedules on. And we have probably increased our forward purchases almost double what we would normally do as shipping windows have stretched from normally somewhere between 60 and 90 days. we actually have one vendor today quoting 15 months for deliveries on key items, another vendor quoting 9 to 12 months. And these were items that historically would ship in 2 to 3 months. So we are also preordering just based on these elongated shipment windows. But I would say the majority of the increase came from price increases to date.

Okay. That's helpful. And then separately on the corporate front, we see from various data providers that leasing in certain metro areas in the U.S. has ticked up quite noticeably as vacancy rates are coming down and whatnot. I'm wondering if you're seeing improving corporate sales trends in those markets? And if so, what that might suggest about corporate growth prospectively?

Yes. So our footprint is heavily concentrated in Class A buildings, which tend to be the first building to recover leasing activity. However, the aggregate vacancy rate and our footprint still remains about triple what it had been historically pre-COVID. So while it is improving, it is improving at a slow pace. Our corporate organic business is growing at around 4% to 5% annually. The decline in corporate has been almost exclusively an off-net and almost exclusively former Sprint customers. Our aggregate Cogent revenue in the 3 years since deal closing has grown at 28%. That results in about an 8% compounded growth rate that has obviously helped by wavelength sales and IPv4 leasing, but we have seen an improvement in corporate on-net growth we have not seen a significant improvement in corporate off-net even for Cogent sales, and we are continuing to see a decline in off-net Sprint corporate as well as an even more accelerated rate of decline in Sprint Enterprise, which is now Cogent Enterprise and is roughly 88% off-net.

And there are no further questions. I will now turn the call back over to Dave Schaeffer for closing remarks.

Hey, thank you all very much. We appreciate everyone taking the interest in Cogent, and we look forward to seeing you at some conferences soon. Take care all. We'll talk soon. Thanks. Bye-bye.

This concludes today's conference call. Thank you for your participation. You may now disconnect.

Great. Let me start by reading a quick disclaimer. For important disclosures, please see the Morgan Stanley research disclosure website at www.morganstanley.com/researchdisclosures. If you have any questions, please reach out to your Morgan Stanley sales representative. I'm Jonathan Greenberg, Investment Banking Managing Director. And I'm here with Dave Schaeffer from Cogent. Dave, thanks for coming back to San Francisco.

Hey, well, Jon, thank you for hosting me. As always, I want to thank Morgan Stanley for a great venue, and I'd like to thank the investors who stayed around to hear a little bit about Cogent.

Yes. Thank you for staying into the evening. I feel like we should all be sharing a couple of cocktails together. But maybe we'll focus on the business to start.

Fair enough.

So let's just start with broader industry trends. Cogent manages 25% of global Internet traffic, which is a pretty impressive number. Traffic is growing around 10% annually. And so how do you think about the surge in demand of traffic and how that influences your business over time?

Yes. So the Internet is roughly 35 years old. And during that period, it's grown at a compounded rate of about 23% a year. So slightly higher than today's current growth rate. But the base keeps getting larger and eventually, it becomes more and more challenging to maintain the same percentage growth rate. I think the Internet actually has decades left of continuous growth. But if we look over that 35-year period, there have been at least 6 discrete technologies that each 1 has adopted has driven a wave of growth.

We are in a transition period from streaming video being the primary driver of traffic growth to now being AI. We're in another 1 of those inflection points. Internet traffic growth is defined by 3 inputs, the number of people using the Internet, the number of minutes a day they use it, and most importantly, the bit intensity per minute. And we have seen streaming growth grow in the past 5 years, starting at the beginning of the pandemic, streaming represented about 18% of all video consumption in the developed world. Today, it's at 34% -- 54%. So we've seen this really S-shaped curve, a very rapid adoption of over-the-top video, professionally produced, displacing linear television. That trend will continue, but we're seeing a maturation of that trend.

On a going-forward basis, we have not yet really seen AI inference be widely integrated in applications and to various business processes. I think as that matures, we will see a reacceleration from the 10% back up probably even above the long-term trend line, which has been the pattern as each new technology gets implemented. And then over a 5- to 10-year period as that technology matures, we start to see things revert back to that [ mid-20% ] type growth rate.

So a surge in demand over this period of time that we've also seen falling prices, which you have historically been very price sensitive in terms of pricing your product. How does that influence your pricing strategy going forward and your positioning in terms of competing for this traffic?

So at an investor conference, I wish I could say that we've reached a point of price stability but that would not be a true statement. Price per bit has fallen also at about 23% per year, keeping that total addressable market in dollar terms flat. I think that price compression will continue perpetually into the future. There was adequate competition both at the end user level and at the backbone level.

Now the real advantage that Cogent has that is sustainable is its network architecture. The core founding principle of Cogent was to build a network focused entirely on the Internet and have the absolute lowest cost of production of an interface-routed bit mile. And when someone uses the Internet, they don't think about what they're doing. They just click on a website and expect something to come back. And what, in fact, a service provider delivers is it gives you an interface to plug into, it routes your bit typically through about 8.5 routers between origin and destination. It typically ultimately traverses a little under 2.5 networks between origin and destination, and travels about 2,800 miles. Ironically, over the 35-year history of the Internet, that has not materially changed. It's isolated some. And what we've seen is 2 factors: more content being cached locally. So therefore, packet travel distances should go down. But offsetting that is the fact that the Internet has become much more global.

So 25 years ago, 85% of all bits on the Internet either originated or terminated in the U.S. Today, that's about 31%. So that globalization has roughly canceled out the localization of content. With all of those engineering data points, how does that translate into pricing and the ability to grow the market size. The 2 underlying technologies that drive the ability to drive down price are wave division multiplexing and optically interface routers. Over that long-term period, those 2 technologies have been on a rapid price performance improvement curve.

Cogent's maybe most important competitive advantage has been its architecture. We capture more of those advances in optronics and in routing more effectively than legacy Internet service providers. We can keep some of that as profit and pass some of that on to customers. So Cogent's market advantage has been the fact that we typically price at a 50% discount to the market average. And that has, in fact, contributed to us being the largest carrier of Internet traffic globally.

So increase in demand falling prices, obviously, the ticket is to take share. Maybe can you talk about -- you just talked about how you take share, but maybe some of the share gains you've seen relative to some of the competitors in the landscape?

So Internet services represent over 85% of Cogent's revenues. We were founded solely on the premise that the Internet would be the only network that mattered. And each of our relevant addressable markets we offer compelling value to the customer. In the end user market where today Cogent has about 1.1 billion square feet of multi-tenant office space, we offer an Internet connection that is typically installed 9x faster than a competitor. It's more reliable once installed, 3x more reliable and then it delivers 30x to 60x the symmetric throughput for price parity. That has allowed us to gain a 35% market share and 11% of all office space in North America.

The bad news for Cogent is that the other 89% of the market is not economically serviceable. The cost to extend fiber to those facilities, the cost of revenue acquisition and the aggregate size of those buildings do not justify the investment to bring fiber into the building and up the riser. So we do have an off-net business where we rely on others to deploy that fiber. But there, our competitive advantage is much lower.

And then for our wholesale or NetCentric business. We connect to 1,902 data centers in 307 markets across 57 countries, far more footprint than any other ISP. We do that over our own fiber network that is comprised of roughly 93,000 route miles of terrestrial fiber intercity and another 33,000 route miles of metropolitan fiber. In that footprint, we connect to what are effectively supermarkets for bandwidth, and we offer our service at a 50% discount. That has allowed us to capture market share. And I think going forward, we continue to see companies who used to compete in that market withdraw. So a decade ago, there were probably 25 legitimate global carriers. Today, there's probably 6 or 7. And eventually, I think this market will stabilize with less than 5 players.

Let's transition to the waves side of the story. This was a key focus of the Sprint transaction that you did a few years ago. You've said that Cogent has less than 2% of that market today, and you think there's opportunity to build that to 25%. Maybe talk about the progress you're seeing in that piece of the business and what it's going to take to get to that 25% target.

So first of all, for investors, let me differentiate between a wavelength and Internet service. Internet service is the connectivity you make when you connect between networks. You use optical transport, but you use routing and interface and interconnectivity to other networks. The Internet is not a single thing, it's an amalgamation of tens of thousands of networks, creating a web of paths that bits can flow across and those bits flow on a routed basis.

The Internet is the easiest to use, cheapest and most ubiquitous connectivity. But there are certain characteristics that customers value that the Internet cannot replicate, security, large package transmission and most importantly, predictable latency with a defined path. That's where a wavelength comes in. We had never been in the wavelength business until 2023. When we had the opportunity to acquire Sprint, we really did 2 concurrent acquisitions at the same time. We acquired an operating business, selling a combination of 3 services. Internet access, MPLS VPNs and a series of noncore products to a number of non wholesale customers, retail customers, either large enterprises or larger midsized businesses. That business that we acquired was burning almost $1 million a day of cash, and we struck a deal with T-Mobile to buy that business and be paid $700 million. And we have ripped cost out of that business, eliminated products, directly shrunk the customer base and turned it into a marginally profitable business.

But the real reason we did the transaction was we got to buy the dormant Sprint LD voice network for $1. We are repurposing that network to sell wavelengths. Sprint had never sold wavelengths, Cogent had not sold wavelengths. We spent over 2 years connecting that network to 1,096 data centers in North America. It's architected solely to deliver wavelengths. We can provision a wavelength between any of those data centers any permutation whatsoever, and there's over 10 to 2,800 power of permutations across that footprint. And we can do that in 30 days or less. That product is much better than what you can buy from legacy providers who sell wavelength services as excess capacity on their network.

The total addressable market for waves globally is about $7 billion, $3.5 billion of that is in North America. The North American market is further split between local, which is dominated by the incumbent phone companies and cable companies that local market is about $1.5 billion, and then a $2 billion intercity data center to data center market. That is what we are focused on. Today, Cogent drives about 4% of its revenues from that business, and we grew 100% year-over-year. And maybe most importantly, we are still a big player in that market.

We have 2% market share, as you pointed out, Jon. And for us to gain market share, we have to go to wavelength customers. It is a relatively concentrated customer base, and we need to show them a compelling value proposition. Why should I use Cogent versus 1 of the 2 or 3 legacy providers? And on that front, we actually have a 5-pronged value proposition. We have more data centers, faster provisioning, unique routes, higher reliability and lower price. A little bit different than in the Internet business, each route is custom priced based on its distance and the wavelength side.

We have been able to use our advantages to quickly gain a foothold in that market, and we need to grow from here. But our competitors are saddled with a network architecture that is less efficient than what we have in delivering those wavelengths.

What should investors take away from this with respect to what you're seeing in the sales pipeline and the current backlog? And you mentioned your time to install is much faster or time to provision. And so how should they think about that translating into results in the near to medium term?

Investors care about 3 things: Can you grow top line? Can you expand margins while you're growing top line and therefore produce an amplified amount of cash flow? And can you have a balance sheet that rationally allocates that cash flow between debt and equity? And I think on all 3 fronts Cogent is addressing those issues. With the acquisition of the Sprint customer base for 9 quarters we had negative top line growth. So for 18 years as a public company before acquiring Sprint, Cogent had a track record of 10.2%, organic growth on average over those 18 years. In the past 9 quarters, that went to negative 5.4% growth. We have now reverted to top line growth.

So what investors need to take away from this is the totality of Cogent is now a growth company. Wavelengths are an important contributor to that. Any of our on-net products carry a similar contribution margin of 90-plus percent incremental EBITDA.

For our office product, we're 35% penetrated. So there's only 65% of the market left on net to capture. And our transit business, we're 25% market penetrated. And in our wavelength business, we're only 2% penetrated. So as the total business grows, waves will be a disproportionate share of on-net growth. With that, we expect Cogent to report a top line growth rate of between 6% and 8% per year on a multiyear glide path.

Secondly, we expect our EBITDA margins to expand because of this relative mix of on-net and off-net traffic by at minimum 200 basis points a year. Before Cogent acquired Sprint, we had an 18-year track record averaging 220 basis points a year of margin expansion. Post Sprint, we've actually averaged over 1,000 basis points a year, but that was coming mostly from cost cutting. Some of it was coming from on-net sales but the majority was coming from rationalizing the burn that we inherited. With most of that rationalization behind us, our growth going forward is going to come from new sales with a high percentage of the sales being on-net.

In the fourth quarter of '25, we had 80% of our sales be on-net. We expect to continue to increase the percentage of the total base that is on-net. Wavelengths will contribute to that. And therefore, we should see EBITDA grow in mid- to low single-digit rates for the foreseeable future. And then finally, we have, I think, taken a number of steps to optimize our balance sheet. The acquisition of Sprint increased our leverage. We had 52 consecutive sequential quarters of growing a dividend. We had to reduce that dividend for the first time because our net leverage was above our comfort zone. We have committed to investors to get back down to 4x net leverage. And at that point, we can begin accelerating a return of capital again to shareholders. I think those are the important takeaways for an investor.

You did a good job covering some of my questions right there. I can come back...

Unless I didn't pique.

How are you thinking about balancing the need to invest in the network for future growth relative to debt reduction you were just talking about? And how do we think about capital allocation as well as return of capital longer term once you do get back there?

So oftentimes, Cogent gets viewed as a serial acquirer. Between 2001 and 2004, we looked at 121 targets. We bid on 19, and bought 13 of them and integrated them into the Cogent network. We ended up paying very little and actually netting cash. We paid $60 million, acquired $115 million of cash, $400 million of acquired preferences, $500 million of debt. We then restructured operationally and financially without a bankruptcy to emerge debt free. And then for 18 years, we did not do a single acquisition. We had looked at 850 targets in past. We chose to do Sprint because it was additive, and it was priced correctly where we felt that the challenge that T-Mobile had gave us the opportunity to derisk the acquisition with the $700 million payment stream. During that period, we returned $2 billion to shareholders. We bought back 22% of our float, and we implemented that dividend and grew that dividend for over 12.5 years successfully every quarter. We did use leverage to do that. And at the same time, we consistently expanded our footprint.

Our business model is very capital efficient. We will need roughly $100 million in long-term average capital per year and another $40 million for principal payments on capital leases. We have debt service that we have to make on the roughly $2 billion of gross debt that we have. And then finally, there's a significant remnant leftover of cash that is available to shareholders. In the short term, we're going to use that along with asset divestitures to accelerate our delevering. But then we will begin to return capital again at an accelerated rate.

We will continue to look at M&A, but I think it's unlikely in this environment that we will do anything and more likely that we will kind of continue to expand our footprint. We're in 57 countries and we consistently add both multi-tenant office space and data centers to the footprint in both existing markets and new markets. The most recent country we've done a significant expansion is in India. That is a regulatorily challenged market. It took us almost a decade to get the requisite licenses. But we do see that as a significant growth opportunity for the company.

With a lot of geopolitical turmoil, it's not clear against this regulatory backdrop, there are that many more countries that welcome an open Internet. We will continue to add. We actually have Thailand on the map to come on-net probably in the next 60 days. So there are still some incremental markets. And then there are incremental cities. So recently, we extended in Japan from just Greater Tokyo to Osaka. So there will be these types of footprint expansions, and we'll typically add about 120 carrier-neutral data centers a year to the footprint. But I don't see M&A as a significant use of capital.

Not a use of capital, but you have been quite vocal about your intentions to monetize the data center portfolio. Maybe can you give us an update as to where those stand?

Yes, sure, Jon. So when we acquired the Sprint network for $1, we effectively got 2 assets, 20,200 miles of dark fiber that was fallow, and 482 buildings that comprise 1.9 million square feet and had 230 megawatts of inbound power. This deal was announced in September of '22. There was no data center shortage problem. There was no constraints on power. Our original thought was we were going to take 45 out of 482, and put a small low power density data center in those buildings. By the beginning of '24, it became clear to us that the asset we had had significant value because of the inbound power. We canvassed a number of potential buyers, listen to what features they wanted to see. And in June of '24, we kicked off the 1 year, $100 million program to retrofit 125 of the 482 buildings. Of that retrofitting process, 24 of the largest buildings are beyond Cogent's ability to fill them up. So we decided to divest them. That project was completed at the end of June of '25.

By beginning of August of '25, 6 weeks after the completion, we actually had our first deal to buy 2 buildings. And those 2 data centers went under LOI, the parties agreed on a price, and in the first quarter was supposed to close. When it came time to close, the counterparty did not renegotiate price but demanded that we provide vendor financing. We withdrew from the LOI, and we pivoted to 1 of our backup deals that people were conducting in the background. Earlier this week, we announced that we have signed a letter of intent. It is nonbinding, but we do believe it will close for 10 data centers and it is substantially higher than the $144 million in the first deal. We'll continue to work with that provider, and we will continue to work with other interested counterparties, some of which are backups to these multiple contracts for the remaining 24 facilities. And we feel comfortable that the quality of the facilities, their location and the aggregate demand backdrop presents us a great opportunity to take an asset that we have no basis in other than the capital we spent to modify them and sold them for a significant gain.

Sticking with the balance sheet for a second. You indicated earlier this week your intent to refinance the senior notes that are due in 2027. Maybe you also published a couple of slides. I think that went up on your website...

I'm going to quiz you on those slides, Jon.

Don't quiz me, let's provide it to the investors right now. But maybe if you could chat through how you're thinking about the refinancing and the structuring that you're doing?

Yes. So let me explain Cogent's organizational structure, which is necessary to understand the refinancing strategy. Cogent Holdings, the public entity has no employees and no debt. Underneath of it, since 2010, there's an operating company, Cogent Group, where all the employees, all the customers and where the debt sits. When we acquire Sprint, those customers went into that entity as well as the subsidy payments from T-Mobile.

We also acquired the MT network that had a $140 million cash carry cost. We did not want to burden our bondholders with that. So we created a sister subsidiary, Cogent Infrastructure, no employees, no operations, just physical assets, the buildings and the fiber. That entity has been funded for the past 2.5 years by the restricted payments capacity generated at the borrower, Cogent Group. Cogent Group has 3 tranches of debt, $623 million of capital or finance leases, $600 million of secured debt, and $750 million of unsecured debt that matures in about 16 months.

And what we are doing is taking the finance leases out of Cogent Group and moving them, leases, liabilities and assets over into Cogent Infrastructure through a series of asset realignment steps. And doing that, we removed $569 million of debt, most senior debt off of the balance sheet of the borrower. We also have committed to put the proceeds from this initial 10 facility data center sale into the borrower group when in fact the assets sit outside of the borrower group. These 2 steps were designed to free up incremental borrowing capacity and enhance the credit of the current bondholders.

We will issue $750 million of most likely 7-year duration, secured debt and use those proceeds dollar for dollar to pay off the unsecured debt, and in doing so, extend our maturities. So the nearest date of maturity will be 6 years, the longest stated would be 7 years. The capital leases would not be in the borrowing group, and they have actually an average remaining life of nearly 21 years in them. So they're extremely long-dated matures.

Lots of initiatives underway, never a dull moment. I would characterize this, though, as a very optimistic conversation. Maybe any parting words for the investors with respect to your excitement for the business and what you're seeing going forward over this year and the years to come?

Yes. I think I have 4 messages to leave investors with. The first, Cogent is a stable business that's been around for 26 years, 21 of them is a public company. We've experienced significant stock volatility and bond volatility over the last year, but our business has been amazingly consistent. Two, we have reverted to top line growth after a 9-quarter period because of the acquisition of the Sprint customer base where we had negative growth. Third, we have demonstrated the ability to expand margins at an extraordinary pace through cost cutting and through product rotation. On a going-forward basis, we will continue to focus on on-net services and grow the profitability of the business. And third (sic) [ fourth ], with the balance sheet tools that we just described, we have an ample runway to ensure that we have ample liquidity to be in a position within a couple of years to be able to return capital again at an accelerated pace to our equity holders.

Great. Very clear. Dave, thanks for coming. We hope we'll see you next year.

Thank you very much. Thanks, everyone, for staying so late. Take care.

Have a good one.

Anyway, once again, welcome to JPMorgan's Credit Conference down in Miami. As most of you know, I'm Tom Egan. I cover telecommunications and technology for the firm. And it's my pleasure today to introduce Dave Schaeffer. If you're in this room, you probably know who Dave is. He's been around a long time. I think I've known him for 25 years. We were both younger men. At least I was, Dave still looks the same. Anyway, Chief Executive Officer and Founder of Cogent.

So Dave, normally, I would start out with talking about the business, and then we'd get into some of the debt-related things like refinancing. But you had some news yesterday at another conference. And so I wanted to make sure that everybody was sort of level set on that stuff. So let's maybe talk about that. As most of you know, during his earnings call, Dave talked about the potential for refinancing the unsecured debt with secured debt. And most of us spent the next couple of days or weeks trying to figure out how you were going to do that. And yesterday, you kind of cleared the air on how that's going to happen. So why don't you take us through what it is you intend to do?

So first of all, Tom, thanks for hosting me. Thank JPMorgan for a great venue. And most importantly, I'd like to thank this full room of investors for taking time to hear about Cogent.

So Cogent has been a high-yield issuer since December of 2010. And our high-yield issuances are done at the Cogent group level. Cogent Holdings is the public company that equity holders own stock in. Underneath Cogent Holdings, there are 2 parallel subsidiaries, one being group that houses all of the operations of Cogent, including the employees, the customers and all of the debt of the company as well as all of the payments that are being made to the company by T-Mobile as part of the consideration for acquiring the Sprint GMG assets from T-Mobile.

To remind investors, we acquired a business that was burning almost $1 million a day. It had a negative 80% EBITDA margin, $300 million of negative EBITDA, $30 million of CapEx, and we were paid $700 million to take that business. When that business was combined with Cogent, we also directed the $700 million subsidy payments to the borrower group. Because those payments were front-end loaded, Cogent actually rapidly delevered. We had historically operated between 3.5x and 4x net leverage between 2010 and 2023. We had grown our equity dividend 52 sequential consecutive quarters.

But with the onset of the pandemic, our leverage creeped up to 4.2x. And then with the acquisition of Sprint and the fact that 50% of those subsidy payments came in the first 12 months, the other remaining 50% were spread out over the next 42 months. Our leverage fell from 4.2x to 2.4x almost immediately. When we acquired Sprint, we did a second acquisition concurrently. We acquired the dormant Sprint network. That was comprised of 20,200 route miles of dark fiber and 482 buildings with 1.9 million square feet of space and 230 megawatts. These were former telephone switch sites.

We got that entire asset base debt-free for $1, but there was a [indiscernible]. There was a negative $140 million burn associated with getting that free asset. An asset that doesn't produce income is really a liability at the end of the day. We had a clear vision of what we wanted to do. We wanted to repurpose that asset to be able to convert it into a wavelength network, and we wanted to convert a portion of the data centers or the buildings into data centers.

We also to mitigate that burn, did place under that silo, Cogent Infrastructure, our asset-backed securitized IPV4 leasing business that has its own ring-fenced sleeve of $380 million of debt. So at the borrower group where the high-yield bondholders have a claim, there were 3 tranches of debt, $623 million of capital lease or finance lease obligations in the form of IRUs, $600 million of secured debt and $750 million of unsecured debt.

Our indentures have 2 limitations on indebtedness, no more than 4x secured leverage, 6x total leverage, inclusive of unsecured and also requires a debt service coverage of 2x. It is an [indiscernible]. But it appeared that we did not have sufficient secured capacity to refinance all of the unsecured with secured. And the way in which we modified the structure enhances the quality of the collateral of bondholders.

So we did 4 discrete things. The first is we took all of those capital leases and the associated liabilities and contributed them into a subsidiary under group. Two, we divided them into the capital leases associated with Western Europe and North America and then the Rest of the World. Those North American and Western European capital leases have a total liability of $569 million. We are selling that subsidiary from group to infrastructure. There is a net cash payment that is coming out of restricted payments.

And as a result, $569 million of the most senior debt disappears from group's balance sheet. In order to be able to continue to use the fiber, Cogent Infrastructure in turn leases that fiber back to group for 10 years. By structuring that as a 10-year lease, under GAAP, it is treated as an operating lease, not as a finance lease.

So the pro forma result is that the EBITDA at group declines by $69 million, the cash payments associated with these leases. On a pro forma basis, inclusive of reclassifying the unsecured as secured dollar-for-dollar paying off the unsecured and putting $750 million of secured in its place, we end up at 3.91x secured leverage under the 4x test and roughly 3x debt service. So we have plenty of room there. We have an additional $100 million of capacity available for us at the secured level that we have no intention of using. And we have about $800 of unsecured capacity that we have no intention of using. Because the fourth point is that at the holding company level, we have made a commitment to all stakeholders that we will dramatically reduce the return of capital to equity until such time as Cogent is at 4x net levered across all of its subsidiaries.

Today, we are at 6.6x. We reduced our dividend by 98%. It went from $1.01 a share to $0.02 a share. It will stay fixed at that level, and we do not intend to have any material equity buybacks until we reach 4x levered. We may do some minor buybacks, but nothing material. With this 4-step program, we have created a dynamic where Cogent Group can easily raise $750 million of secured debt and that debt would sit pari passu with the existing 6.5% secured debt that matures in 2032. And temporarily, it will be extended 1 additional year.

So we want to make sure that the new bonds have a maturity date at least 1 year beyond the current bonds. In order to further enhance the collateral pool of the borrower, we voluntarily said that we would commit the proceeds from our pending LOI for the sale of 10 data centers. We have no obligation to do that. Those assets sit at Cogent Infrastructure outside of the borrower group. But in order to further strengthen the credit, we felt that it made sense to inject that capital. Quite honestly, it was sleeves out of our vest because we already had committed to 4x net leverage.

Now I have been asked today, does that commitment guarantee that we immediately purchase debt? Or can we just leave the cash on the borrower's balance sheet. And we're going to preserve that flexibility, but the cash will be trapped inside of Cogent Group. It is subject to removal via restricted payments tests. We do use some restricted payments in order to fund the operating burn at the infrastructure level, which is still existing, but we intend to continue to be able to grow EBITDA.

Last year, we grew our underlying EBITDA for the full year, $70 million. Our EBITDA margins expanded year-over-year by 800 basis points. While we expect EBITDA growth and margin expansion going forward, I do not believe 800 basis points a year is sustainable, but we do believe that at least 200 basis points a year over a multiyear period is consistent with what Cogent has done historically and is prospectively possible. We also have attributed most of the low margin or negative margin acquired Sprint revenue. We've turned that operating business from burning $300 million to today slightly with probably low single-digit type EBITDA margins.

We still have more work to do. We still have more cost to cut. But most importantly, the aggregate combined Cogent has returned to top line growth. For an 18-year period as a public company with no M&A, Cogent organically grew at a compounded rate of 10.2% a year. That growth rate turned negative 5.4% when we acquired the shrinking business from Sprint. Now that, that business has been purged of undesirable revenue, the combined company is continuing to grow. The organic Cogent business during that 9-quarter period grew 27%. The Sprint business, which represented 42% of revenues of the combined company at closing, shrunk by 64% during that same period. That rate of revenue attrition has moderated, and we're now in a position where the aggregate business is growing and margins are expanding. Hopefully, that was clear enough, Tom.

I think so, but I'm going to summarize it anyway because that was an awful lot to digest. So let me just see if I've got it right. Cogent's fiber that is under IRUs where you the fiber via IRU, you're going to take $569 million of those leases, which are now senior to the $750 million senior secured, and you're going to take them off the balance sheet and you're going to move them over to the unrestricted group. And then you're going to pay the unrestricted group $69 million in lease payments for those -- the use of that, and it will be a shorter-term lease.

So it will end up not being a capital lease, not being on the balance sheet, it will end up being an operating lease. And then you're going to take whatever proceeds you get from the sale of the data centers, you're going to migrate that cash back into the restricted group. And when you're done with that, that's what gets you to the 3.91x net leverage? Or is that even before you do the money from the data centers?

It is before we do the money from the data centers. So the data center sale is a nice to have, but not a must-have. So we will most likely do the financing prior to closing the data center transaction, and it is not necessary. The only point I would clarify, Tom, is that the capital leases actually set senior not only to the $750 million of unsecured group, they sat senior to the $600 million of existing secured.

That's actually what I meant -- that's what I said.

So now the secured debt, which will grow to $1.35 billion will be effectively the most senior. There will be $54 million stub capital leases in some less developed countries.

Right. Okay. Got it. And is the $69 million in payments that you're going to make for the operating leases, is that the same amount of money that you were paying under the IRU when it was on your balance sheet? Or is that a...

No markup, no profit. Dollar for dollar. So there's no transference of value, and it's just meant as a mechanism to inject the capital into infrastructure so it can meet the capital lease obligations.

Okay. And normally, we'd continue. But what I want to do is because this is such an important topic to debt holders, I want to make sure that if there's anybody in the audience that has any questions on this topic. We'll move into something shortly. But if it's on this topic, could you please raise your hand and ask Dave?

And do you mind giving this gentleman maybe a mic so that people on the webcast can hear it and maybe I can hear it too then. Can you speak up?

[indiscernible]

Is that a traditional new issue and then the cash proceeds take out the unsecureds? Or is it you offer a secured security to unsecureds for them to tender into?

No, this is not an exchange offer. The current unsecureds mature in June of '27. They have a make-whole that lasts until June of '26. If we do a transaction between now and June of '26, we will fund into escrow and then break that escrow when the make-whole reverts to 0. But this is not an exchange offer. If we do the transaction after June, it will just be a takeout at par of the unsecureds.

Yes, sir.

Just a question on the structure. Given the proceeds from any potential data sale, I understand there's a new LOI. Maybe you could just give some clarity on that. But committing the proceeds to the issuer group, because those are from the legacy Sprint network, the new waves business that you're building off that, does that sit in Cogent infrastructure?

So I'd say somewhat complicated question. Let me answer the 2 pieces of it. First of all, the data center sales are for physical assets that exist off of the Sprint infrastructure and are owned by Cogent Infrastructure and the proceeds that will come from a nonbinding LOI from a major global infrastructure fund for 10 facilities that is substantially more than $144 million. Those are the facts that are in the public domain. Those proceeds are being pledged to be not retained by infrastructure, but to be immediately contributed to Cogent Group to enhance the credit portfolio.

The second part of the question, which is the building of the wavelength business. The Cogent infrastructure entity had already granted an IRU from infrastructure to group for 8 fibers across the entire footprint. That allowed Cogent Group to abandon a Lumen IRU and save $15 million a year in expenses. That was well documented in previous public disclosures. And it also allowed us to build a parallel new wavelength network off of completely fallow fiber.

The equipment we use to do that is all owned by group. The equipment came from 3 sources: some new equipment purchased, some older Sprint legacy equipment that was just sitting dormant that we reactivated and then some older Cogent IP transport equipment that was then diverted to the wavelength network. So the wavelength business and all of the revenues associated with it go directly into group. The burn associated with the empty network and the empty buildings sit in infrastructure, and it is why that we use about $100 million a year of RP capacity from group to fund that negative burn at infrastructure.

And just a follow-up. In the new LOI, the 2 data centers that were part of the previous LOI that failed due to last-minute renegotiation of terms, are those 2 previous data centers in this current package for potential sale?

On e of the 2 is. And that was actually the critical gating item for this potential buyer. They were looking at the asset when we signed the original LOI for the first 2, but we're not ready to move to an agreement. And we chose the deal that was ready to go for 2 facilities.

When those 2 were withdrawn, that party continued the work on the remaining 9 that they were interested in, but they were very clear to us they would not proceed unless they could get that 10th facility. When the original LOI party came back to us and demanded financing, it gave us the ability to withdraw from that LOI, which is what we did. And then we were free to enter into negotiations with the second acquirer for all 10 facilities, and we were able to conclude that in a matter of less than 2 weeks under terms that were acceptable to both parties.

That buyer has done physical tours in 8 of the 10 facilities as of last week with a third-party consultant. They've had access to the data room, and they are completing their last 2 site visits this week. Their primary gating item is confirmatory due diligence on the power availability at the sites. These sites were fully powered to 109 megawatts in 2015.

In 2015, Sprint decommissioned its TDM voice network. Since that, for the remaining almost 11 years, these sites have been really just drawing minimal power on a keep alive basis. Before we decided to convert those facilities, we went to each utility and did our own confirmatory due diligence to see that the power is available. There are no PPAs or purchase power agreements. All of this power is delivered at tariff rates.

We have e-mails and confirmations from the engineering department that if we were willing to commit to the utilization, the power is available. The counterparty is validating the confirmations we got. I'm sure they've got additional due diligence. They're doing Phase 1 environmentals and they're doing title abstracts. We had done the title abstracts at acquisition. The Phase 1s, we relied on those that T-Mobile did when they acquired Sprint.

Anything else on this topic?

We have another one over here on the left.

I'm just curious how competitive the process was with this LOI. Do you have other counterparties behind them for other data centers, these data centers, et cetera?

So just to refresh everyone's memory on the time line, we acquired Sprint. We announced the initial tranche of data centers in May -- or actually June of '24, we began conversion in earnest on 125 facilities, including these 24 large facilities. At the end of June of '25, we ran a process and we chose the winner out of probably a half a dozen LOIs that were on the table.

At that point, and we chose that winner because of both price and their ability to confirm availability of funds. The other party, including the one that ultimately concluded was doing work but had not progressed far enough to put in a firm offer. We have continued to build a pool of what I would consider backup LOIs and probably have today, again, somewhere between 5 and 10 of those types of agreements. But many of them, I think, are with parties that we don't have confidence will get to closure based on funding requirements.

We've had literally dozens of counterparties trying to tie up the assets in exclusivity, owing to then go out and shop them to raise capital. And our view was before we would tie an asset up, we needed to know that the party tying it up could, in fact, perform if we delivered what we said we would deliver.

Okay. That actually cleared up a whole bunch of questions I had on data centers. Maybe just one more, David, was -- if my memory serves me correctly, you had originally something like 23 data centers that you considered to be superfluous...

24.

24 data centers that were superfluous. What are you thinking about the remaining 14, let's say?

We think that there will be buyers for some. Our hope is to sell all them. Each one is slightly different. They're in different geographies, different power. They're all of about the same vintage and design. They were all built as telephone central offices to house switchgear. The total of 482 facilities, most of those, about 355 of them, we concluded are just not suitable for conversion. They're just too remote or too small or both.

Of the remaining, we have actually converted now 125. Most of those, 101 of them are small, and we think we can fill them up with our traditional colo model of selling 1 or 2 racks at a time. These larger facilities, 24 of them are of a different scale. The fundamental difference was the smaller facilities hold Class 5 switches, the larger facilities held Class 4 switches. There's probably nobody in this room or even old enough to remember what the difference in those 2 switch types are because they don't make any of them anymore. They're all in the scrap heap. But the bigger sites were designed mostly for long distance and the smaller sites were more for regional or local traffic.

I might end up on that scrap heat because I remember those switches. One of the questions I get quite frequently is regarding your waves business. You had originally given us a target of I think it was $500 million?

$500 million by run rate by midyear '28.

By midyear '28. And at this point, that looks like a stretch. Now maybe, you're...

We only did $40 million last year.

I understand that. But it still looks like a bit of a stretch. Could you just talk a little bit about your confidence in that number? Or -- and maybe just a little bit about besides the things that you've talked about before where folks make the order and then you put it together and then they don't want it turned on for 6 months because they weren't expecting it to be on in the first place. Could you just talk a little bit about maybe what you might have missed on getting to that number quicker than maybe not getting to that number as quick as maybe you thought you would have?

So first of all, I think we can still achieve the multiyear target that we laid out. That target was laid out in September of '22. It was meant to be 5 years post closing, so 5 years mid-'23. And I think we needed to lay out some justification. Sprint was not in the wavelength business. Cogent was not in the wavelength business. What we know is there is a total addressable market for North American intercity waves of about $2 billion.

We also built out the most ubiquitous footprint. We initially targeted 800 data centers. That's actually grown to 1,096 as of the end of fourth quarter. That is a much broader footprint. We have 5 discrete competitive advantages over the current wavelength providers. We have a broader footprint, faster delivery, unique routes, higher reliability and lower prices. Those characteristics should allow us to capture 25% of the market.

Our Wave business initially was manually provisioned between 65 data centers, a subset of the 800. In 18 months, we actually installed about 1,000 waves in that footprint. But we also knew that model would not scale and would not get us to our endpoints. We were busily working on all of the steps necessary to take a TDM voice network and turn it into a wave network. We completed that work in the end of 2024 -- December 31, 2024. Over the next year, the Wave business grew year-over-year 100%. It grew sequentially in the fourth quarter, 19%.

If we could keep that growth rate up over a larger base, we can, in fact, hit our numbers. We need to continue to earn credibility in that market. The fact that we've delivered waves to 518 sites and to about 200 unique customers is helping us build that credibility.

The market is dominated by companies that fail to deliver. And when they do deliver, it takes anywhere from 3 to 4 months to do so. We are attempting to change that customer expectation. We are making real progress. Tom is correct. We've installed waves in advance of customers being ready to take them. That gap is starting to close. We also probably made a mistake being too granular around both funnel size and exact progression in a specific product.

Cogent does not give quarterly guidance for any of its metrics. It was necessary in the short term while we were building that business. We will maintain that granularity of disclosure historically. But on a going-forward basis, the metrics that investors should hold us to are top line growth of all products of somewhere between 6% and 8% over a multiyear period measured annually, an average of 200 basis points, not including the 800 that we did historically, but just going forward. And then third, maintaining a debt maturity that will ensure that we have adequate runway and liquidity to be successful.

We want to not deemphasize waves. Waves are, in fact, the easiest way for us to grow because we only have 2% market share. In the transit business, we have 25% of global market share. In our multi-tenant office building business, where many of your firms are customers, we have 35% market share. It's always easier to grow from a small share to a large share, and that's why waves are critical. But any on-net product carries identical contribution margins of over 90%.

Fantastic. Dave, we've run out of time, but appreciate you coming here.

And I didn't get through hardly any of those questions, Tom.

You didn't on that. Didn't I ask too. Thank you.

Thanks, Tom. Thank you.

Good morning, and welcome to the Cogent Communications Holdings Fourth Quarter and Full Year 2025 Earnings Conference Call. As a reminder, this conference call is being recorded, and it will be available for replay at www.cogentco.com. A transcript of this conference call will be posted on Cogent's website when it becomes available. Cogent's summary of financial and operational results attached to its press release can be downloaded from the Cogent website.

I would now like to turn the call over to Mr. Dave Schaeffer, Chairman and Chief Executive Officer of Cogent Communications Holdings. Please go ahead.

Thank you, and good morning to everyone. Welcome to our fourth quarter 2025 and Full Year 2025 Conference Call. I'm Dave Schaeffer, Cogent's CEO, and with me on today's call is Tad Weed, our Chief Financial Officer. I'd like to highlight a few key events and significant matters in the quarter. I'd like to be able to go through these metrics to help you understand better, our business.

We are continuing to increase our margins. Our increase in gross margin and EBITDA margins have been driven by cost reductions and a rotation to more profitable on-net products. In the third quarter of 2023, the first full quarter Cogent was combined with Sprint Wireline revenues, our combined revenues by connection type for the third quarter versus this quarter have changed materially. Our on-net revenues were 47% of our revenues in the third quarter of 2023. Our total -- our net revenues as a percentage of revenues has increased from 47% of revenues in the third quarter of 2023 to 61% of revenues this quarter. Our off-net revenues were 48% of our total revenues in the third quarter of 2023, immediately after the combination of Sprint and Cogent. Our off-net revenues as a percentage of our total revenues have decreased from 48% of revenues down to 39% of total revenues this quarter. And our noncore revenues were 5% of total revenues in the third quarter of 2023. And our noncore revenues as a percentage of our total revenues has decreased to less than 1% of our revenues this quarter.

We'd like to take a moment and outline our progress in our wavelength sales. At year-end, we were offering wavelength services in 1,068 locations, all capable of 10 gigabit, 100 gigabit and 400 gigabit services with provisioning intervals of approximately 30 days. As of today, we have actually increased our service footprint to 1,096 locations. Our wavelength revenue for the quarter was $12.1 million, a 74% year-over-year increase compared to the comparable quarter in 2024. Our sequential wavelength revenue growth accelerated and increased by 19%. That is better than the 12% sequential increase in Q3 over Q2. Our wavelength customers increased by 18% sequentially to 2,064 connections at the end of the quarter.

Our wavelength revenue for the full year 2025, which was the first full year we were selling wavelength services across our footprint, was $38.5 million, an increase of 100% from the 2024 number. Our wavelength customers during that period increased by 85%. As of the end of the quarter, we have sold wavelengths in 518 locations compared to 454 locations at the end of Q3. We continue to anticipate capturing 25% of the highly concentrated wavelength market in North America.

Now for a few comments on margins. Our EBITDA as adjusted for the quarter increased by $3 million to $76.7 million. Our EBITDA as adjusted margins for the quarter increased sequentially by 140 basis points to 31.9%. Our increased margins continue to come from our cost reductions, as well as our product optimization. Our EBITDA as adjusted for the full year 2025 was $55.6 million. Our EBITDA as adjusted, then adding back the payments under the T-Mobile Transit Agreement. Our decrease in EBITDA as adjusted was as a result of the $104.2 million reduction in our IP Transit payments from T-Mobile and a reduction of $21.4 million for other reimbursable Sprint acquisition costs that we incurred in 2024. There were no Sprint acquisition costs in full year 2025. The $104.2 million reduction in scheduled payments and $21.4 million reduction in these acquisition costs more than offset the organic growth of $70 million in Cogent's EBITDA or EBITDA classic for full year 2025.

Our EBITDA classic for 2025 was $192.8 million. For the full year of 2024, it was $122.8 million. Our EBITDA as adjusted margins were 30% for the full year 2025, down from the 33.6% for the full year 2024 because of the reductions that I just previously mentioned. Our EBITDA classic margins, however, for full year 2025 were 19.8%, up from 11.9% for full year 2024, or an improvement of approximately 840 basis points on a year-over-year basis.

Under our IP Transfer Agreement with T-Mobile, we will continue to receive an additional 23 monthly payments of $8.3 million per month until November of 2027. There are further cash payments related to lease obligations we assumed at closing of at minimum, $28 million. This $28 million payment is to be made by T-Mobile in 4 equal monthly payments from December of 2027 through March of 2028.

Now for a comment on our improvement and leverage. We have refined our capital allocation priorities and strengthened our financial flexibility and accelerated our delevering strategy. Our leverage ratios have improved. Our gross debt leverage as adjusted for amounts due from T-Mobile for the last 12 months EBITDA as adjusted ratio was 7.35% as compared to 7.45% in the previous quarter. Our net debt ratio was 6.64% in Q4 compared to 6.65% in Q3 of 2025.

We believe that the amounts due from T-Mobile under our Transit and Purchase Agreement should be considered and calculated our leverage ratios. We believe that these amounts essentially represent both long-term and short-term cash and are discounted appropriately on our balance sheet. And due to T-Mobile's credit rating and payment history, we are confident that these payments will be -- continue to be made in a timely manner. T-Mobile pays us $25 million a quarter through the fourth quarter of 2027 under this IP Services Agreement. The monthly payments from T-Mobile under the IP Transit Agreement reduces from the balances that are due each month as they are received.

Now for a couple of comments on our improved IPv4 leasing activity. Our IPv4 leasing revenue increased 44% year-over-year to $64.5 million for full year 2025. We are currently leasing 15.3 million addresses at year-end. This is an increase of 2.2 million incremental addresses or 17% on a year-over-year basis. We have titled to 37.8 million IPv4 addresses.

Our capital expenditures for the last half of 2025, once our data center modernization program had been completed, was $73.3 million as compared to $114.3 million for the first half of 2025. This $41 million decrease was due to the completion of a significant amount of reconfiguration work in our Sprint acquired facilities. We have converted these facilities into data centers in the first 6 months of 2025, as well as the last 6 months of 2024. We have converted a total of 125 facilities. At year-end, we are providing facilities in -- providing services in 1,715 carrier-neutral data centers as well as the 187 Cogent data centers. The Cogent data centers have an aggregate capacity of 213 megawatts of installed and available power.

Now as many of you know, we have intended to monetize and sell 24 of these facilities that we view as surplus. We acquired these facilities through the acquisition of Sprint, and we intend to monetize them through either outright sale or leasing on a wholesale basis. The nonbinding letter of intent we mentioned on our last call was not finalized due to a change in the original terms, not in price, but a requirement by the purchaser for Cogent to provide a portion of the purchase price in terms of owner financing, which we found unacceptable. We reverted to some of our backup agreements and are in active discussions with multiple parties for multiple offers across a broad set of these data centers. We do expect several of these to result in multisite acquisition offers.

Now for a moment about our leverage and balance sheet strategy. Our 2027 June unsecured notes of $750 million are still roughly 18 months from maturity, but we have begun receiving proposals to refinance these notes. We intend to complete a refinancing transaction for new secured notes of $750 million as soon as the make whole period expires in June of this year.

Now for our long-term goals. We anticipate our revenue growth to continue to improve and be in the 6% to 8% range. We expect our rate of EBITDA margin to actually moderate to the roughly 200 basis points a year that we will be able to deliver over a multiyear period. The nearly 800 basis points that we delivered this year was due to some extraordinary cost savings. And while we will continue to deliver these results, we do expect the rate of margin expansion to moderate. Our revenue and EBITDA guidance are meant to be multiyear goals and not intended to either be quarterly or even annual guidance.

Now I'd like to turn the call over to Tad to provide some further detail and provide our safe harbor language. Tad will also give a further breakout of the trends in the revenues acquired from the Sprint base versus the Cogent classic base since our acquisition in 2023. I know this has been an area of focus of investors, and we have been able to disaggregate those revenues and now present them and with clear trends and metrics. With that, we'll then open the call up for questions and answers. Tad?

Thank you, Dave, and good morning to everyone. This earnings conference call includes forward-looking statements. These forward-looking statements are based upon our current intent, belief and expectations. These forward-looking statements and all other statements that may be made on this call that are not historical facts are subject to a number of risks and uncertainties, and actual results may differ materially. Please refer to our SEC filings for more information on the factors that could cause actual results to differ. Cogent undertakes no obligation to update or revise our forward-looking statements. If we use non-GAAP financial measures during this call, you will find these reconciled to the corresponding GAAP measurement in our earnings releases that are posted on our website at cogentco.com.

Some overall comments on results and revenues. Our total revenue for the quarter was $240.5 million and $975.8 million for the year. Our total revenue for the quarter declined sequentially by $1.4 million or by 0.6%. This was an improvement from the $4.3 million or 1.7% sequential quarterly revenue decline that we experienced last quarter. While our sequential revenue declined within our fourth quarter, our total revenue increased each month in the quarter. Our total monthly revenue increased from September to October, increased from October to November, and excluding the change in USF revenues, increased from November to December. This month-to-month total revenue increase continued from December 2025 to January 2026.

There was a negative FX impact on our quarter [ sequentially ] revenues of $0.2 million. So for the quarter, we experienced a $2.2 million sequential decline in off-net revenues. Our on-net revenues, including on-net weight revenues, increased by $0.9 million or 0.6%. And our noncore revenues decreased by $0.2 million, and now those revenues have declined to only $1.2 million. Sequential wavelength revenue growth, which is on-net, accelerated to 18.8% from 12.4% last quarter and increased sequentially by $1.9 million.

Gross margin. Our gross margin for the quarter increased sequentially by $1.6 million to $112.5 million. Our gross margin increased sequentially by 100 basis points to 46.8% from continued cost reduction and product optimization, including our focus on our on-net products. Our gross margin for full year 2025 increased by $46.7 million to $442.7 million. And our gross margin for full year 2025 increased by 720 basis points from 38.2% last year to 45.4% for full year 2025.

EBITDA. Our EBITDA, not including payments under the IP Transit Agreement for the quarter, increased sequentially by $3 million to $51.7 million, and our EBITDA margin increased by 130 basis points to 21.5%. Our EBITDA for the full year not including the IP transit agreement or Sprint acquisition costs increased by $70 million to $192.8 million from $122.8 million for full year 2024. And the EBITDA margin for this year increased by 790 basis points from 11.9% to 19.8% for full year 2025.

We analyze and classify our revenues into 4 network connection types and 3 customer types. Our 4 network connection types are on-net, off-net, wavelength and noncore. And our 3 customer types are NetCentric customers, corporate customers and enterprise customers. Dave mentioned we'll provide some information on Sprint Wireline acquired revenue and Cogent classic revenue. We have been hesitant to separately disclose our revenue performance related to our acquired Sprint Wireline business and our Cogent classic business once the operations have been fully integrated. However, we believe that the following analysis will be beneficial in understanding some of the changes in our total combined revenues.

The substantial changes in the acquired Sprint Wireline revenue base have masked the underlying performance of our legacy Cogent classic business. So in May 2023, when we closed the transaction, the Sprint Wireline revenue base had a run rate of $39.4 million per month or $118 million per quarter. This acquired revenue base has decreased from that $118 million per quarter at the acquisition date to down to $43 million for this quarter. That's a $75 million quarterly revenue decline related to the Sprint Wireline revenue base or a 64% decline since the deal closed.

At deal closing, our Cogent classic revenue run rate was $155 million per quarter. This quarterly revenue base has increased by 27% or by $42 million from that $155 million prior to close to $197 million for this quarter, the fourth quarter of 2025. Additionally, our Cogent classic revenues increased sequentially by 1.5% from the third quarter of this year, increased year-over-year by 3.1% from the fourth quarter of 2024 and increased by 2.3% for full year 2025 over full year 2024. Our consolidated revenue declines have been largely attributed to the reduction in the acquired corporate and enterprise revenues from Sprint. At closing, the Sprint Wireline revenues represented a total of 42% of our total revenues, and that percentage has materially dropped from 42% down to only 18% of our total revenue at year-end. Our total corporate business was 42.7% of our revenues this quarter and 43.9% for the year. Our quarterly corporate revenues decreased by 9.1% year-over-year and sequentially by 2.3%. For the year, our total corporate revenues declined by 9.7%.

At the closing of our acquisition of Sprint Wireline in May 2023, the Sprint Wireline corporate revenues were 30% of our total revenues. Those Sprint Wireline acquired corporate customers now represent only 10% of our total corporate revenues. The Sprint Wireline acquired corporate revenue base has decreased from a run rate of $13 million per month or $39 million per quarter at closing to a run rate of $2.7 million per month or $8.1 million per quarter at year-end 2025.

The same analysis for NetCentric. Our total NetCentric business continues to increase and benefit from the growth in video traffic, activity related to artificial intelligence, streaming, IPv4 leasing and wavelength sales. Our NetCentric business was 43% of our revenues this quarter and 40.3% for the year. Our quarterly NetCentric revenues increased by 10.4% year-over-year and sequentially by 3.1%. For the year, our total NetCentric revenues increased by 6.8%.

At the closing of our acquisition of Sprint Wireline, the Sprint Wireline NetCentric customers represented 20% of our total NetCentric revenues. Those Sprint Wireline acquired NetCentric customers now are representing only 7% of our total NetCentric revenues this quarter. The Sprint Wireline acquired NetCentric customer revenue base has decreased from a run rate of $6 million per month or $18 million per quarter at closing to a current run rate of $2.9 million per month or $8.7 million per quarter at year-end 2025.

Lastly, the enterprise business. Our total enterprise business was 14.3% of our revenues this quarter and 15.8% of our revenues for the year. Our quarterly enterprise revenue decreased by 24.7% year-over-year and sequentially by 5.8% primarily due to reduction in the acquired noncore enterprise and off-net low-margin enterprise revenues. For the year, total enterprise revenues declined by 20.3%. At the closing of our acquisition, the Sprint Wireline enterprise customers represented virtually 100% of our enterprise revenues, as this was a new line of customer for Cogent. The Sprint Wireline acquired enterprise revenue base has decreased from a run rate of $20 million per month or $60 million per quarter at closing to a current run rate of $8.8 million per month or $26.4 million per quarter at year-end 2025. These substantial changes in the acquired Wireline revenue base have masked the underlying performance of our legacy Cogent classic business.

Analysis on revenue by customer connection network type. On-net revenue. We serve our on-net customers in 3,579 total on-net buildings. For the year, we increased our on-net buildings by a total of 126 on-net buildings, similar to prior years. Our total on-net revenue, including on-net wave revenues, was $146.4 million for the quarter, a year-over-year increase of 7.8% and a sequential increase of 0.6%. Our total on-net revenues, including on-net wavelength revenues, increased as a percentage of our total revenue by 400 basis points to 58.4% for this year from 54.4% for full year 2024.

Off-net revenue. Our low-margin off-net revenue was $92.9 million for the quarter. That was a year-over-year decrease of 17.9% and a sequential decrease of 2.3%. Our off-net revenue results are impacted by the migration of certain off-net customers to on-net and the continued grooming and termination of acquired low-margin off-net contracts. Our total off-net revenues decreased to 40.7% of our revenues for this year from 43.8% for full year 2024.

Some comments on pricing. Our average price per megabit for our installed base decreased sequentially by 12% to $0.14 and by 34% year-over-year, essentially in line with historical trends. Our average price per megabit for our new customer contracts was $0.06, a sequential decline of 18% and 46% year-over-year.

ARPUs for the quarter. Our on-net IP ARPU was $509. Our off-net IP ARPU was $1,234. Our wavelength ARPU was 2,114. Our IPv4 ARPU was $0.30 per address. Churn rates. Our churn rates improved sequentially. Our on-net and off-net churn rates improved from last quarter. Our on-net unit monthly churn rate this quarter was 1.2% compared to 1.3% last quarter. Our off-net unit monthly churn rate was 1.9% compared to 2.1% last quarter. And our wavelength monthly churn rate has been less than 0.5% to relatively insignificant.

Traffic. Our year-over-year IP network traffic growth accelerated for the quarter. Our IP network traffic for the quarter increased sequentially by 4% and by 10% year-over-year. And for the total year, our traffic increased by 9%.

Sales rep productivity. Our sales rep productivity was 4.1 units this quarter compared to 4.6 last quarter and 3.5 in the fourth quarter of 2024. That's compared to our long-term sales rep productivity average of 4.8.

Foreign currency. Our revenue earned outside of the United States was about 20% of our revenues this quarter, similar to prior quarters. Based upon the average euro and Canadian conversion rate so far this quarter, so the first quarter of 2026, we estimate that the FX conversion impact on [ sequential ] revenues will be positive about $0.4 million. And year-over-year, more significant, about $3.5 million.

Customer concentration. Our revenue and customer base is not highly concentrated. Our top 25 customers are 17% of our revenues this quarter, similar to prior quarters. CapEx. Our CapEx was $37 million this quarter and $187.6 million for the year. And principal payments on capital leases were $8.5 million for the quarter and $33.8 million for the year. Combined, those amounts have declined year-over-year.

Comments on debt and debt ratios. Our total gross debt at par, including $623.4 million of finance lease obligations under long-term IRUs, was $2.4 billion at year-end. Our net debt -- total net debt of our cash and our $203.1 million due from T-Mobile at year-end was $1.9 billion. Our leverage ratio, as calculated under our more restrictive covenants under our unsecured $750 million 2027 notes indenture, was 6.13. The secured leverage ratio was 3.8, and the fixed coverage ratio was 2.38. The definition of consolidated cash flow, similar to EBITDA under our $600 million secured 2032 notes indenture includes cash payments under our IP Transit Services Agreement with T-Mobile and the definition and determination of consolidated cash flow.

Payments under our IP Transit Agreement were $100 million for the last 12 months, so that is added to the calculation. Our leverage ratio, as calculated under the $600 million secured 2032 notes indenture, was 4.67. Our secured leverage ratio was 2.9. And lastly, fixed coverage was 3.12. Bad debt and days sales. Our days sales outstanding was 30 days at year-end, the same as last quarter. And our bad debt expense was less than 1% of our revenues for the quarter and for the year.

And with that, I will turn the call back over to Dave.

Thanks, Tad. I'd like to highlight a few of the strengths of our network, our customer base and our sales force. Now for some details around our NetCentric performance. We continue to be a direct beneficiary of a number of trends in the industry, whether it be artificial intelligence or streaming activity. At year-end, we're able to sell wavelength services in 1,068 data centers across North America with a provisioning interval of approximately 30 days. At year-end, we're selling IP services globally in 57 countries and 1,902 data centers. At year-end, we were directly connected to 7,659 networks. That is the largest number of directly connected networks of any service provider on the Internet. 22 of these were peers, and the remaining 7,637 networks were, in fact, Cogent Transit customers.

Now for some details around our sales force. We remain focused on sales force productivity and are disciplined about managing out underperformers. Our sales force turnover rate was 5.4% a month in the quarter, down from a peak turnover rate of 8.7% during the height of the pandemic and also below our historical average turnover rate of 5.7% of the sales force per month. At year-end, we had a total of 590 quota-bearing reps. Our sales force included 289 sales professionals focused entirely on the NetCentric market, 289 sales professionals focused on the corporate market, and finally, 12 sales professionals focused on the enterprise market.

In summary, we have made significant progress in a number of areas. We've improved our revenue trajectory and performance and have returned to sequential revenue growth, which we expect to continue. We are improving our margins and growing our EBITDA due to our diligence and cost reduction and our focus in selling profitable on-net services. Over 80% of our sales in the fourth quarter of 2025 were for on-net services.

We have a clear plan to refinance our 2027 $750 million unsecured notes with a new, longer duration $750 million secured note offering. We are actively working to monetize some of the acquired Sprint facilities, which will further accelerate our delevering and allow us to resume a more aggressive return of capital program to our equity holders.

We are effectively have now completed the integration of Sprint and Cogent's network into a unified network and business. We have converted all of the intended Sprint switch sites that we intend to convert into data centers. This program is materially complete and will result and a continued reduction in our capital intensity.

We are enthusiastic and optimistic about our wavelength business to add to our product portfolio. Our wavelength services are differentiated due to the uniqueness of the routes, the breadth of our footprint, our efficient provisioning and aggressive pricing. The reliability that we deliver is unparalleled. We have, since inception, offered superior services, a broad footprint of revenue-rich locations, expedited provisioning and market-leading disruptive pricing. That is why Cogent continues to be a market leader in the products that we sell.

With that, I'd like to open the floor up for questions.

[Operator Instructions] Our first question comes from the line of Chris Schoell with UBS.

Dave, you had previously talked about returning to sequential revenue growth while sustaining sequential EBITDA growth each quarter. Can you just update us how you're thinking about total company revenues and EBITDA for 2026 as that Sprint mix continues to fall? And as we think about the waves business scaling in 2026, any guardrails you can share on the number of connections or revenues you believe are achievable based on what you're seeing in the business right now?

Yes. Thanks for the question, Chris. So as we mentioned, we are not in the habit of giving specific quarterly or annual guidance. But I do believe that after the significant runoff in the Sprint acquired revenues, as Tad pointed out, 64% of the revenues that we acquired 2.5 years ago have [ attrited ]. And during that period, the Cogent revenues, which represented 57% of the combined company, had grown at 27%. As a result of that, we have had now about 10 sequential quarters of revenue growth. We will be back to positive revenue growth on a quarterly basis from this point forward, and we anticipate that the annual rate of growth on average over a multiyear period will be in that 6% to 8% range. We also have a small amount of further cost reductions that will contribute to margin expansion.

But the primary driver of margin expansion going forward will continue to be the revenue mix shift and the focus on on-net services. 80% of our sales in the quarter were on-net. We have improved the base from 47% on-net immediately after closing to 61%. We think that percentage will continue to improve and allow us to achieve that, at minimum, 200 basis point rate of margin expansion. The reality is we did nearly 800 basis points last year. That is probably not sustainable over a multiyear period, but we do have some tailwinds there.

And then to your question around wave. We have the largest North American wave footprint. We are beginning to gain credibility with customers. We saw an acceleration in our revenue recognition and installations. We expect those trends to continue. And because our wavelength products are virtually all on-net, they are significant contributors to our margin expansion.

Another way to kind of look at the markets that we operate in. In our on-net multi-tenant footprint, we today have about a 35% market share. That means we can continue to grow there, but it is harder because we already have over 1/3 of the customers as Cogent customers. In the NetCentric market for IP services, we are the largest provider globally and have 25% market share. We will continue to gain share and grow that business. But again, with 25% share, it becomes incrementally more difficult.

And what's encouraging to us about wavelength is the fact that we have less than 2% market share in North America. We are now establishing our credibility with 518 sites now having actual reference customers in them. And nearly 1,100 sites wave-enabled, we think that our rate of wavelength growth will accelerate and help us drive that kind of 80:20 mix in the incremental business.

Our next question comes from the line of Gregory Williams with TD Cowen.

Sam on for Greg Williams. Two, if I may. First on the waves business, you mentioned before that the goal was to get the funnel to the waves funnel to about 10,000. Is the idea to get the funnel to 10,000 and it kind of stays in that range because you install the backlog as it comes in? Or do you expect the funnel to grow from there?

And second, on data centers. You mentioned the contract changes that pushed out the LOI for the 2 data center assets mentioned on the 3Q call. Is the expectation this transaction will still close? And if so, is the $144 million a taxable event? Or is there some sort of tax shield from the Sprint deal?

Yes. Let me take those in reverse order. On the LOI that we announced last quarter, the counterpart, it came back to us and look for us to provide more than 50% of the agreed to purchase price in owner financing. Since we had a number of other interested parties who had submitted backup offers on those 2 facilities as well as a broader set of facilities, we decided to terminate that agreement at our choice and then reengage with some of those parties. We are far along in those negotiations and hope to be able to announce something soon. And that announcement may be for a broader set of assets. Now to the tax consequences, I'll let Tad touch on that.

Sure. So as a reminder, we paid only $1 for the Sprint business. So the tax basis is essentially the assumed liabilities, which is minimal in the -- both the buildings and the network that was acquired. However, we have material NOLs this year from the tax bill from 2025. And given the bonus depreciation deductions, we expect to continue to incur tax losses to offset any gain on the buildings going forward. So why it is a taxable event creating taxable income, we don't think on a net basis, that will result in income taxes being paid.

And now, Sam, I'll touch on your waves question. While we were in the process of enabling the footprint, we felt it was critical to give funnel KPIs to show expressions of interest by customers. We have tried to be clear with investors that we do not give funnel data routinely for our other products, and we're treating wavelengths now as any other product. Now we do in our investor presentation typically show both our on-net and off-net conversion rates for the previous quarter. We intend to continue to do that. Our funnel is continuing to grow, but we will not be reporting specific numbers. But we do anticipate with the footprint that we now have and the credibility that we are earning with existing customers, we are starting to see a larger percentage of their wave opportunities being shown to us for bid. And as a result of that, we will close more and see further acceleration in the waves business.

Our next question comes from the line of Sebastiano Petti with JPMorgan.

Just a quick follow-up on the waves business there. Could you update us on the level of installed but not yet billed balance in the quarter? Did that grow off of the third quarter? Because I think last earnings, you probably talked about maybe a few hundred waves had been installed, but not yet billed. And so what is the progress there? And then I have a follow-up.

Yes. So 2 points. First of all, in the quarter, we actually saw the unit number of waves improve, which is an indication that we were eating into that backlog. But we also have been building an additional backlog. And I would say that the installed but not yet billed base is comparable this quarter to where it was at the end of third quarter.

Got it. That's helpful. And then I guess, maybe just help us think about back to the data centers to some extent. I mean, I think you did mention that there were some other data centers that had been in active discussions last quarter. And so while the LOI that you just spoke of on the third quarter that's kind of now been terminated, what was the progress on some of the other, I guess, remaining data centers that were in active discussions? Did those progress? And I guess, maybe help us think about as you look at your debt refinancing and the stack later this year, I mean, yes, you talked about trying to perhaps refinance with $750 million of secured. I mean, is there some level of assumed cash proceeds from asset sales anticipated in the intervening period as well, which probably helps maybe reduce the prevailing interest rate you might get at that time?

Yes. So really, three different questions. The first one is some of the backup offers on the 2 facilities that we had mentioned previously cover those facilities and others. So some parties were not particularly interested in moving forward without those facilities potentially being included. So it was not a 1 for 1, meaning that there was a backup just for the 2 facilities that were under LOI. And our view was that while there was no difference of opinion on price, we felt that we would be better served with an all-cash purchase rather than one that had us taking more than 50% of the purchase price in the form of a secured note against the assets.

In our refinancing, we are not assuming that there will be proceeds from the data center sales, although I do think there will be some proceeds. They are not baked into the point that I've made around the timing of the refinancing. Our plan is to refinance the unsecured notes with secured notes, dollar for dollar, no increase or decrease, in aggregate phase value and do that in a way that allows us to avoid paying the make whole, which would be due in -- any time between now and mid June of about $13 million.

The final point I will make on that is that the proceeds for the data center sale would be reflected as cash on our entire balance sheet, but the proceeds do not go into Cogent Group, which is the borrower group of both the secured and unsecured debt. We may elect to contribute some of that cash to group, but we're well within the coverage ratios, both in terms of secured and total indebtedness and also in debt service coverage. So there's no requirement for us to contribute that capital, but it would be available at an unrestricted sister entity, Cogent Infrastructure, and therefore, could be used to either inject that capital into Cogent Group, the borrower or dividend back to Cogent Holdings, which can then be used for the benefit of shareholders.

Our next question comes from the line of Frank Louthan with Raymond James.

Great. So on the data center, I think you'd originally kind of focused on $9 million or $10 million per megawatt. I mean, what do you think the market is for that now? And why not maybe try and lease those out and then get a multiple on that value? And then what additional room do you have on pricing and maybe leasing additional IPv4 licenses?

Frank, let me take those again in reverse order. In terms of IPv4 leasing, we saw a material acceleration in our leasing, but at lower price as we did 2 wholesale transactions of large blocks. We are continuing both on a retail and wholesale strategy. Today, we're about 46% of our addresses are leased and approximately 4% of our addresses are allocated to customers at no cost. This is nothing new. It's been part of our strategy to win business since Cogent's inception.

But we do still have half of our address space that is sitting fallow. We have greatly improved the marketability of that address space by being able to deploy RPKI or additional security features across those addresses, which have made them more desirable to counterparties. And we anticipate continuing to see growth in our IPv4 leasing business. The 44% year-over-year growth in that business, again, was extraordinary. I'm not sure if we can repeat that, but we will continue to see further growth.

Out to the data centers. I think when we established a go-to-market strategy in the spring of '24 and announced that we were going to begin the capital investment to convert these facilities, we looked at both public trading comps as well as transactions in the private market. If anything, over the past year, data center space has become more scarce and valuations have improved. Now we are fully conscious of the fact that our data centers are repurposed switch sites and not purpose-built campuses which are different and attract a different customer base. We have done a minimal amount of leasing and have been focused mostly on the sales process. I think we feel that based on the number of sites that are in active discussion and a number of counterparties, that we will absolutely be monetizing through sale, a significant portion of the footprint.

And in terms of exact price per megawatt, we are not going to disclose that because that would impact our ability to maximize value through these negotiations. But as Tad pointed out, other than the capital that we've invested, we have no real basis in these assets. And in fact, because the assets sit at Cogent Infrastructure, they represent a negative EBITDA cost that's not burdening the borrower, Cogent Group, but is a drag on the entire complex. And by selling these data centers, we both get the cash proceeds as well as a reduction in operating expenses.

Our next question comes from the line of Brandon Nispel with KeyBanc Capital Markets.

I appreciate the analysis on the Sprint revenue versus Cogent Classic. I wanted to understand and ask a few questions there. First, maybe just can you help us understand how you came up with that analysis? Because I think in the past, you've said it's sort of difficult or impossible to delineate between the 2 businesses once you integrated.

Second, what changed versus your expectations? I think, Dave, when you closed that acquisition, you said you'd probably be at more of a run rate of $350 million versus $190 million annualized run rate that you gave us today? And then where do you think the bottom is, what do you think that business does in terms of revenue in 2026?

I'll take those again in reverse order. One, I think that business is continuing to deteriorate, both based on the nature of the customers and the discipline that we have applied to ensuring that the services we sell have an adequate margin. While we realized that the off-net enterprise customer base is inherently less profitable, in fact, even after a diligent effort of trying to bring enterprise business on-net, we have only been able to get to an 88% off-net and 12% on-net mix because many of these enterprises operate globally across a footprint that is just not economic to bring on-net. And therefore, we're going to be saddled with that lower margin portion of our revenue stream, but we do intend to make sure that the margins are adequate.

We have virtually completed the burn-off of the noncore products and the vast majority of the undesirable revenue. But with that said, we're still experiencing significant monthly and quarterly sequential degradation in that business. I had projected the 10.9% rate of decline that we were seeing from Sprint. We thought that we could maintain that rate of decline and migrate customers to more profitable products. What we in fact found was that many of those customers actually intended to go away independent of our acquisition at an accelerating rate. And then that was further compounded by the discipline that we applied. I think it will continue to decline. We will continue to report on it.

Now in terms of why we did this and how, it was a very arduous and manual task. We had to go into the nearly 1,300 acquired customers and look at every individual order on an order-by-order basis. It was a very manual process. But I do believe, based on concerns I was hearing from investors, that this was an extremely important metric that they cared about. And we then basically invested what was effectively a full-time person to do this analysis. We will be able to do this going forward. And I think it gives an investor a better lens on how Cogent's business is performing versus the acquired business, as well as the mix shift that we are focused on and being more on-net.

The way to improve our cash flow going forward is growth in top line, but growth in top line of more profitable business. And the 80% on-net that we sold in Q4 is actually better than we did the quarter before we acquired Sprint. So in Q1 of 2023, we actually only were 76% on and 24% off. So this focus on on-net is going to be a significant driver of margin expansion.

I'll just add one thing to the complexity. So when we acquired the business under the TSA, T-Mobile was billing the customers on our behalf through their billing system. We worked an incredible effort to bring all of those customers into our billing systems. We had 1 billing system for November 2023. But for that period, from close, so May 2023 through October '23, we were relying on the information that we got from T-Mobile billing on our behalf. So bifurcating that and post billing on our system was complicated, I'll just leave at that.

Understood. And if I could just follow up with one quick one. Where would you estimate the EBITDA contribution of the Sprint business is today?

I think it's close to 0, but slightly positive. But still far below our aggregate margins. It's probably in the 0% to 5% range. But we are working on improving that, which does include, in some cases, price increases.

The next question comes from the line of Nick Del Deo with MoffetNathanson.

A couple of questions on the data center front. Dave, you were explicit that the LOI fell apart because of the demand to help you finance it. I recall that one of the due diligence items that the counterparty needed to complete was confirming power availability from the utilities. Was that availability confirmed?

It was confirmed by that party. And we have now made the data available to the backup providers to go through the same confirmation process. But the reason why we did not move forward with the previous LOI was not a negotiation on price. They got comfortable with both the power availability and title to the actual land, which were their 2 big concerns. And they just came back and tried to have us provide them financing. Even though when we executed the LOI, they had assured us that they have proof of funds and the wherewithal to pay all cash. They were just trying to magnify their returns through owner financing.

Got it. Are you able to share when the LOI fell apart? And if you do have new deals in hand soon, as you suggested, should we expect a press release to announce those? Or would you disclose those on your next earnings call or some other conference presentation or something?

I think we would probably announce it in a stand-alone announcement. And I do think we anticipate something in the next couple of months. That's probably as specific as I can be. But unless it was a day or 2 before the earnings call, I think we would probably announce it separately.

And then to when the LOI fell apart, it was fairly recent. There was a negotiation. They had made the request, we went back, and we're trying to keep them moving forward under the original terms. But eventually, earlier this year, we became convinced that they were not going to move forward unless we provide the financing.

Okay. Okay. And then can I ask a couple about the legacy Cogent versus legacy Sprint revenue splits. So it looks like you're talking about a $42 million growth in quarterly legacy Cogent revenue from the time of the deal closing to today. It looks like over that time, your quarterly IPv4 revenue is up about $9 million. Waves are now at about $12 million. So that would imply that about half the revenue growth was from those 2 line items and about half came from, call it, the core products that you focused on pre-deal. Is that a fair way to think about it? And...

Yes.

Is it correct to assume that the -- okay. Okay. Good. And the T-Mobile TSA revenue, is that in the Sprint bucket?

No, it is not. That was revenue that did not exist previously and was a drag to our revenue. I guess it was about $400,000 for the quarter -- last quarter. And I think at peak, it was almost $6 million. So that was the services we were providing to T-Mobile that we had previously never provided. And it was not to a Sprint customer, it was to T-Mobile. But they have been able to reduce their reliance on our paid services by about 93%, 94%. But that remaining $400,000 is in there. So in fact, the underlying Cogent revenue growth probably would have been a little better if we had excluded the TSA both initially and today.

Okay. Got it. That's helpful. And if I can squeeze in one more quick one about the IPv4 leasing revenue. So the revenue was down a little bit quarter-over-quarter despite the address at least being up noticeably. Can you just talk about the dynamics there?

It was actually pretty simple. One of the parties that took the large wholesale block had a small retail block with us, and it was the timing of when we terminated that retail agreement and converted it to wholesale in conjunction with a much larger purchase.

Our next question comes from the line of Michael Rollins with Citi.

Dave, I was curious if you could be more specific on the cost base. I think in the past, you described that there's some duplicative costs that the company is incurring during this integration. How much of those are left, and the timing of those savings? And then can you also share with us what the burn rate is for the data center portfolio that you're looking to monetize?

Yes. Two very different questions. So first of all, we have achieved the vast majority of the increased cost savings that we had targeted. So if you remember, the initial number was $220 million. We then increased that number to $240 million. And we probably have achieved over $230 million of that $240 million in cost savings. So there is a small tail, but it is not material.

Secondly, we have incurred incremental expenses associated with integration activities. Those will continue throughout this year. They peaked at about an annual run rate of $60 million or about $5 million a month. Today, they're down to probably closer to $3 million a month, but there is still monies being spent on various integration optimization programs, but we do anticipate those ending by the end of the year.

And then to the final question, which is the burn associated with the infrastructure that we acquired from T-Mobile. So the infrastructure business, which includes the data centers and the physical fiber network, has a negative EBITDA of about $140 million. We have partially offset that because the IPv4 securitization sits under infrastructure and generates about $60 million of EBITDA. So the infrastructure silo of Cogent's balance sheet is about negative $80 million of EBITDA. Roughly 20% of that is associated with the data centers, and we're looking to sell a significant portion of that footprint, probably at least 50% of it.

And sorry, that 20% associated with the data centers, is that 20% of the...

20% of the $140 million of negative costs associated with the Sprint assets. These are primarily in 3 buckets. They're real estate taxes, personal property taxes and right-of-way fees. We got the actual network for $1 with no revenue. We now are completing the repurposing of that. And as we add high-margin -- the margins accrued [ to ] group, but we can fund those losses over at infrastructure through our ability to move money out of the borrower group through holding some back down to infrastructure. In fact, that's how we've been funding those to date using our restricted payments capacity. And we do have about $350 million of accumulated unused restricted payments capacity at the borrower.

If I could just follow up real quick with two other items. First, if you look at the corporate business at the heritage Cogent side of the equation, can you share with us a little bit more detail about what's driving the heritage revenue change over the last couple of years? And if there's any inflection in trend there? And the same for NetCentric, where it might be a little bit easier to unpack the IPv4 and the wave impact, just given the concentration of those products in NetCentric?

Yes. So on the NetCentric side, it is easier because we do break out the IPv4 revenue, of which 85% of it is NetCentric. We break out the wave revenue, which is virtually all NetCentric. And then the incremental difference is the growth in the core transit product.

In the corporate business, there was a mix of DIA and VPN services at Cogent and then a mix at Sprint. At Sprint, the mix was much heavier VPN than it was DIA. At Cogent, it was much more DIA. We have converted some of the Sprint customers from MPLS to VPLS VPNs, improving the profitability, but we are continuing to support the MPLS product long term. We are trying to move as much on-net as possible. But the underlying growth in the corporate business at Cogent has come mostly from DIA.

Our next question comes from the line of David Barden with New Street Research.

The first one, Dave -- and I apologize for asking this -- is about your new contracts that you've signed in January with the Board and how we, as investors from the outside, look at maybe how your incentives have changed. You always took stock as compensation. Now you're getting cash compensation. Does that change how you think about the business, how you think about dividends? It would be really helpful to get some insight there.

I think the second question, maybe for Tad, is when you talk about secured financing, what specifically are you planning on securing? How much are you planning to secure? And what rates are you expecting?

All right. Great. So first of all, with regard to my contract, I am still in negotiations with the comp committee for some additional equity going forward. The vast majority of my compensation, roughly 80% of it remains in equity, and that equity does not begin the vest until 2029. So there's both a long-term cliff and a significant portion that needs to vest. Now -- so I do not have to pledge shares going forward, which created a cascade of bad events. I now have cash compensation that will allow me to pay both taxes and to be able to live, but it is a fraction of my total compensation.

In terms of being able to go forward and how I think about dividends, I'm as committed to shareholder returns as I've ever been. We have shifted our priorities to get our leverage down. And I think we will be in a position where we will see our leverage rapidly fall and be able to return to either buybacks, dividends at a higher rate or a combination thereof. I'll let Tad touch on the refi, and I may jump in as well.

Well, I mean, we're in negotiations with multiple parties. We've essentially only kind of come to terms on the amount, but not with respect to rates and the rest of the terms that we are in the process of negotiating.

Yes. I think we have a very clear structure that will allow us to do this as secured debt. I don't think this call is the correct forum to roll that out, but we will see. And we also will anticipate that the current secured debt is a pretty good education of about where our new debt will price.

Got it. And is there anything about the 2032s that is relevant to kind of rolling the '27s?

Not really. I mean, the same test will be in place will be governed by the most restrictive covenants, which will probably be the existing '32s, and that will be 4x secured leverage and a 2x debt service test.

Got it. And if I could just squeeze in one more, I really appreciate it, guys. Dave, you've kind of mentioned that the two kind of things that were going to be advantages for you in the wave market were price and time provision. I think you said you're down to 30 days, I think you targeted 2 weeks. Could you elaborate a little bit on the kind of process to get to even better provisioning timing? And where are you, do you believe, on a price perspective relative to market?

Yes, I'll take the price one first. I think we're probably at a 20% to 30% discount. I also believe our advantages are more than you outlined. I think the breadth of the footprint, as well as the diversity of the routes and reliability, are all really important criteria. And I think the acceleration you're seeing in our waves business is as a result of that.

And then in terms of getting the provisioning window even shorter, I think it will be three things. It will be, one, [ our just ] continued process refine it as we do more, but 30 days is still generally 3x to 4x quicker than industry averages. A third party actually just last month, released a report benchmarking us. And in terms of wavelength services, out of all of the providers -- there were several dozen providers, both regional and national, evaluated -- we were actually #2 in terms of provisioning already. And I think we'll end up being #1, just like we are in IP.

I think the other thing that is a constraint today is actually, pluggable optics lead times have become more challenging just because of the pressures that some of the massive data center builds have put on the entire ecosystem for telecom and networking equipment.

Our next question comes from the line of Mike Funk with Bank of America.

I've had one question, Dave. Going back to the sequential revenue growth, I'm looking at the Street forecast, and consensus is for about $3.5 million sequential revenue growth in 2026. And this is not '26-ish. I think historically, straight forecast revenue growth faster than actual probably a combination of constructive commentary from the company. The longer-term revenue growth guidance provided and relative opaqueness of your business. I don't think it's helpful to have revenue growth so much higher than actual. So maybe help us think about the correct rate of sequential revenue growth in 2026 to reduce some of the volatility that we see in your stock on earnings?

Yes. And it's a delicate balancing act because while I want to give clarity and guidance, I'm not comfortable in giving quarterly or even annual guidance. I do think that over a multiyear period, that 6% to 8% growth rate is what is absolutely appropriate to model. I'm going to have to leave it to every analyst to do their own diligence and channel checks, and we're just not going to give a number that says 3.5 is too high or too low on a sequential increase in revenue.

What we said is, from this point forward, we're comfortable that our quarterly reported revenues are going to grow. We think that's going to continue to improve. We think that, that growth is going to be driven by high-margin products. And just as you said, maybe Street numbers were too high on top line, they've consistently underestimated our ability to expand margins.

Maybe one more if I could, Dave, sneak it in here. Rep productivity, I wanted to touch on that. They've been coming down. What are you doing internally, change in processes, people, to improve rep productivity?

So the productivity is measured on a unit basis. If you've actually noticed, our ARPUs have actually gone up somewhat too. We are focused more on on-net services. So there is a higher payout for on-net versus off-net to help get to that 80:20 mix that I referenced. And then third, we continue to train, to promote internally and try to incent our sales force to grow. But we do still have 5.4% per month of turnover. That is below the long-term average of 5.7%, the peak of 8.7%. But our productivity at [ 4.1 ] for the fourth quarter was actually about 18% better than our rep productivity in the fourth quarter of 2024.

There is some seasonality to rep productivity. And while the 4.8 that we average is good, we actually think we can do better than that. And I think you'll see that number trend up as this focus on on-net and as we kind of roll through the seasonality that I mentioned.

Our last question comes from the line of Ana Goshko with Bank of America.

So just on the plan to refi the '27 sort of dollar for dollar with new secured. So in the prior question on the planned use of proceeds of any data center sale, you clearly didn't commit to using it to repay debt. So I think you said you have options. But when you reduced your dividend about -- in the last earnings announced date, the rationale that was provided for reducing the dividend was that you wanted to focus on deleveraging. And I think implicit in that was the idea that cash on the balance sheet, potential cash from asset sale proceeds and potential cash from free cash flow would be used to repay debt. So I just wanted to revisit that concept and what the plan is to get leverage down? And I believe you cited a target of 4x?

Yes, that is absolutely correct, Ana. And we are absolutely committed to not materially changing our return of capital either through buybacks or dividends until we reach 4x net leverage. We each quarter have less money due to us from T-Mobile, which we're counting in our leverage. So that is a bit of a hill that we have to climb. We also are again, delevering both on a gross and net basis. And I think we will continue to do that.

And holding cash on the balance sheet has the exact same impact on net debt. We have not been specific around a gross debt target, and we may opportunistically even buy back some of our debt if it's some [ concentrates ] at a discount as our current secured debt has, that could also be an effective mechanism to use excess cash to reduce leverage. But we are absolutely committed, I don't want to leave no ambiguity, that we intend to get for the entire complex, not just the borrower group, down to 4x net leverage before we materially change our return of capital strategy.

Okay. And then secondly, I know you don't provide specific guidance. But in terms of your ability to generate free cash flow, particularly this year, what is your level of confidence and maybe some order of magnitude if you expect it to be positive?

So we absolutely will produce a growth in EBITDA. You can extrapolate what we have done and then later on, the contribution margins with the mix shift that I described and then layer in some of the aggregate savings. Two, we absolutely expect our capital expenditures to go down. Those two things will allow us to generate unlevered free cash flow growth. And it is likely that when we refinance the unsecureds, our coupon will be slightly higher probably than it is today for the current unsecureds even though we will be converting them to secured. That is highly dependent on how the current bonds trade. But we do think that even on a levered basis, we will be generating free cash. That's as close to guidance as you're going to get me.

And that concludes our question-and-answer session. I will now turn the call back over to Mr. Dave Schaeffer for closing remarks.

Well, first of all, I want to thank everyone. I know it was 1.5 hours. We have actually gone longer. I thought this was somewhat unique in that we added a lot more granularity to our disclosures around the trajectory of this acquired business and also, the relative mix of products.

I think in summary, there are 3 really important objectives for Cogent to build value. One is to grow top line. Two, to continue to expand margins. And then three, eliminate any overhang of a debt maturity that is 17 months away. And I think on all 3 of those vectors, we are and will continue to demonstrate meaningful progress. Thanks, everyone, and we'll talk soon. Take care. Bye-bye.

This concludes today's conference call. You may now disconnect.

All right. I think we'll get started. Hello everyone. My name is Chris Schoell. I'm on the Communications and Media Research team here at UBS. And today, we're pleased to have Dave Schaeffer Chief Executive Officer of Cogent Communications here with us. So Dave, thank you for joining us. .

Hey, Chris, thanks for hosting me. I'd like to thank UBS for a great venue, and I'd like to thank all the investors and the audience for taking some time to hear what we're up to.

Just quickly before we get started, I have to read a quick disclosure statement. So as a research analyst, I'm required to provide certain disclosures relating to the nature of my own relationship and that UBS with any company on which I express a view today. The disclosures are available at www.ubs.com/disclosures.

Alternatively, please reach out to me, and I can provide them to you after the presentation. So Dave, given the time of the year, maybe you can start off by talking about what you saw as the key developments over the past year for Cogent and the priorities as you look out into 2026.

So Cogent it's definitely been a mixed past year. I think we've made some tremendous strides in terms of expanding our wave network and converting the assets that we acquired from Sprint into assets that can now be monetized. We took the former Sprint long distance voice network and converted it into a network to sell optical transport or waves. That network now serves over 1,000 data centers, and we can provision a wave from any data center to any data center in 30 days or less.

We have converted 125 of the former switch sites into data centers. We have earmarked 24 of those facilities with 109 megawatts of inbound power for sale to divest of them and unlock value. Our core business has actually continued to improve and is probably growing slightly faster than it was prior to the acquisition.

Now on the negative side, the business that we acquired from Sprint that is selling Internet access and MPLS VPNs to corporate end users had been previously declining at 10.9% a year for the 3 years prior to the acquisition, and post acquisition in the 9 quarters that we have reported, it has declined at 24.2% year-over-year. That accelerated rate of decline was intentional.

We wanted to groom noncore products. We wanted to migrate off-net services to on-net and for the remaining off-net services that we keep, we wanted to improve margins. During that period, we have grown our EBITDA absent the subsidy payments from T-Mobile from about 1% margins back up to 20%.

So prior to the acquisition, our margins were 40.5% in the quarter immediately preceding the transaction. Our margins fell in the quarter immediately after the transaction to 1% and are now back at slightly above 20%. We have received -- over half of our $700 million subsidy payment from T-Mobile for taking over this money losing and declining business.

But due to the capital expenditures and the operating losses, our aggregate leverage has increased to 6.6x. As a result of that increase in leverage we have decided to reduce our dividend by 98%. We've taken a fair amount of pain for that. Now we do expect our underlying EBITDA to continue to grow, and our margins from this point forward to continue to expand at about 200 basis points a year.

That's a great overview. Maybe starting with the wave business, I think last earnings call, you talked about you still believe by 2028, you can get to this $500 million revenue goal for the wave business, which would be roughly 25% of the market. It does imply that, that business needs to scale faster than it is today. What gives you confidence that, that can be done and what still needs to be done to accelerate the core trends there?

So just to reiterate, the total North American wave business is about $3.5 billion. The intercity portion of that is $2 billion, and our goal is to be at 25% market share run rate by midyear 2028. That implies a $500 million revenue run rate and our current revenue run rate is about $40 million after the reporting of third quarter. So in the initial period from acquisition in May of '23, through December of 2024, we were working on repositioning that network.

We initially configured 65 data centers where we can provide wavelengths. We were successful in selling about 1,000 wavelengths in that limited footprint that is less than we had originally anticipated. Since the beginning of the year, we've effectively grown that business in 3 quarters by 80% and are doing about $10 million a quarter, $10.1 million in Q3.

That revenue number has to accelerate. We have seen a building of our sales funnel, a faster rate of installs and an improvement in the rate at which customers are accepting service. That needs to accelerate further. We will win market share based on 3 factors: One, we're lucky. The market is growing because we had originally anticipated a market of either regional networks, international carriers or content delivery and did not anticipate the incremental demand from AI, which has allowed the market to grow.

Secondly, we have a superior value proposition. What does that exactly mean? Most consumers equate value with price. And that is an important component, but there are multiple dimensions which we offer a better service than our competitors. We're in more data centers, 1,000 versus our competitors that are in 300 or less.

Two, we can provision more quickly. Three, we have unique routes in 90% of the instances. That's extremely important for a service that is unprotected, and therefore, having diversity is critical to reliability. Fourth, because we acquired these assets for $1 they had a cost basis of $20.5 billion, and T-Mobile sold the physical network to us for $1.

We have the flexibility to price more aggressively. Today, we're pricing at about a 20% discount to market on a route-by-route basis. And then finally, we have greater reliability. We have greater reliability because of the physical installation of the fiber that we own -- it's buried deep beneath railroad tracks in an armored cable as opposed to plastic conduit near the surface near public highway right of way.

As a result, it's had less cuts per kilometer, and we'll have less cuts going forward. So I think those 5 competitive advantages, coupled with the fact that the aggregate market is growing should easily allow us to hit our goals.

And we've seen some of your main competitors like Lumen, AT&T announced faster provisioning time lines. They're expanding their own wave networks, do you worry that this takes away some of your competitive edge when you go to market?

So one should always be worried about their competitors, you can never be complacent. In the case of those 2 companies, there were a number of asterisks in those announcements, limiting the number of sites, limiting the paths between those sites. The network that we have built is solely designed to carry wavelengths.

It can support an any-to-any configuration which represents over 10 to the 2,500 power number of permutations. It is impossible to preprovision those many routes to be able to have capacity waiting for the customer. You do it in response to the customer. We took a very different approach to the market than our competitors. We are not selling remnant capacity off of a multipurpose network, but rather we built a network from the ground up solely to deliver wavelengths.

We are the largest carrier of Internet traffic in the world. We have built a separate network in order to deliver IP transit in more data centers, nearly 1,900 data centers in 58 countries. We provisioned faster than anyone else, and we price at half of the competitive market price. That has allowed Cogent to dominate that market and become much larger than AT&T, lumen in that space.

I think those same lessons will be translated into our wavelength business. You win business one customer at a time. And for each of those customers, you just need to deliver a better value proposition than your competitors and you win. We can't manufacture demand. What we can do is offer a better solution to those customers who are demanding the service.

And one of the challenges you cited was it was taking time for customers to adapt to this faster provisioning. Have you seen much improvement on that front here in 4Q?

Chris, we've seen a slight improvement. But -- we came to a market where the norm from legacy providers was 3 to 4 months to install with a 50% failure rate on installs, meaning they took an order and the order was never provisioned for the customer. We came to market saying, one, we'll do it in 30 days. Two, we'll do it across this footprint of 1,000 sites.

The market rightly so, did not believe the claims we made and we're slow to accept the services that we are delivering. As we have now delivered services in 500 locations, and we have delivered services to over 200 unique customers, we are beginning to build credibility. Cogent has a reputation of being an IP transit provider. We are now in the process of building that reputation as a wavelength provider. And I would suspect over the next several quarters, that time from install to customer acceptance is going to continue to shrink.

And we've seen more headlines about these AI-related power investments in more remote locations. Can you just help frame for us, is that an opportunity for your wave business -- and I think in the past on some of these greenfield network builds, you've kind of downplayed the returns that maybe some of your peers were seeing on these? Has your calculus there changed at all?

So, we have made a conscious decision not to deploy capital into single-tenant location, because we would then be beholden to that single customer, whether it's a corporate endpoint or a purpose-built data center. However, we have 3 ways in which we can serve those locations. One, when companies are building proprietary data centers in a greenfield, they will typically secure dark fiber back to a carrier-neutral location.

At that location, they will then interconnect to transcontinental networks such as our own, and we will be able to sell services using a combination of customer supply dark fiber and our own and our city network. The second case would be where we go out and buy dark fiber to go into one of these greenfield builds. We will only do that if we combine the dark fiber under a contract term that's matched to the customer commitment. What we don't want to do is find ourselves with a stranded asset.

And then the third permutation would be to buy lit services, effectively selling a compounded wavelength that uses a combination of our own network and another network. Our competitors do that routinely. The margins to create and the fact that we're in over 1,000 data centers means that we will do it less. But it is likely that if there are a single tenant locations that require waves, we will end up using some off-net techniques to get to those locations.

And maybe pivoting over to your corporate business. So the declines there have been elevated as you have gone through the grooming process with -- on the Sprint assets, which I believe those revenues are either low or almost 0 margin. How much of that is still left to work through? And when can we see that piece of the business return to sequential revenue growth?

So in our acquisition of Sprint's business from T-Mobile. There were actually 2 separate transactions that just happen to occur at the same time. The first was, we took over a declining business, selling a combination of noncore products, MPLS services and Internet access services of which 93% of those services never touched the Sprint network. They were off-net in their entirety.

We have groomed the noncore products by 90%, taking the run rate from about $60 million a year down to below $6 million. Our ultimate goal is to get that to 0 as quickly as possible as those are negative gross margin services. Secondly, we are selling access services, whether it be for MPLS or DIA off-net.

The enterprise space, which is entirely acquired from Sprint, is today 88% off-net, 12% on that carries low margins, and we have identified a number of locations that were not viable. We will only deliver off-net services using fiber. We don't want to use fixed wireless coax or twisted pair as it is unreliable and does not have adequate throughput. We also identified locations and markets that we are not licensed. Cogent operates as physical network in 58 countries around the world. In every one of those countries, we have a license to sell Internet and VPN services.

We recently entered India as a new market, that was our 58th market with those types of licenses. We have been grooming away Sprint business in markets where we do not have a license, some sub-Saharan African countries, some Central Asian countries are good examples of this.

As a result, our enterprise business and our corporate business has declined, and it's primarily been an off-net services. That off-net rate of decline was $7.1 million sequentially last quarter. That was a peak as we jettisoned the majority of that remaining business. We would expect that our enterprise business will continue to decline at a couple of percent a year.

We would expect that the acquired Sprint corporate business will decline also at a couple of percent a year, but the corporate business that is organic Cogent will probably grow at around 6% to 8% a year, returning that total corporate business to growth. It's also important to remember the mix of on-net versus off-net. So prior to the acquisition, in the last quarter, we reported in Q1 of '23, Cogent had 76% of its revenues on-net, 24% off-net, and our EBITDA margins were 40.5%. After we acquired Sprint, margins fell to 1% and -- the mix fell to 47% on-net and 53% off.

In the 9 quarters since the acquisition, we have returned to 61% on and 39% off. It has been this rotation from off-net to on-net and the grooming of these unprofitable services that has allowed us to grow EBITDA for 9 sequential quarters while top line has declined on average 2.4% a year.

And you mentioned the legacy corporate business growing 6% to 8%, driving total corporate back to growth. What does the time line look like for that? .

Total revenue growth at Cogent, which includes NetCentric and enterprise will probably be this fourth quarter and the corporate specifically we may be a quarter or 2 behind as our NetCentric business is growing faster, and our enterprise business is certain to decline at a couple of percent. Enterprise today represents roughly 15% of total revenues.

And then maybe moving over to NetCentric. I think traffic growth is the main driver in that business. It's been around 8% to 9% in recent quarters. It used to be at a double-digit clip. Is it fair to say that this level of traffic growth persist until you see AI be more of an uplift? How are you thinking about that trajectory? .

So our revenue growth has actually been above historic averages even though traffic growth has been below historic averages because of the internationalization of traffic. So today, roughly 55% of our traffic is ex-U.S., whereas 20 years ago, less than 15% of traffic was outside of the U.S. We typically get higher revenue per bit in international markets, which is helpful to us. And we have seen aggregate Internet traffic growth slow to about 7% from a historical average of about 22%. Cogent's traffic growth has slowed to about 9% against a historical average of about 25%.

So while we are still gaining share, we're doing so at a moderate rate. Is difficult because we are already the largest provider in the world with a quarter of global traffic. But the drivers of growth have historically changed over time, initially e-mail, file transfers, casual video, professional video, and now we're at the cusp of AI driving another leg of global Internet traffic growth.

So AI exists because of the Internet. The roughly 1,000 zettabytes of data that have been collected and stored is the wrong material for building large language models and AI training. The influence that will come out of those models will be distributed around the world and accessed via the Internet, not via closed networks. And as AI gets more integrated into applications, you will see aggregate big usage per minute go up and number of minutes of usage go up. so just as streaming movies drove a decade of Internet traffic growth. We're now at the cost of AI driving that growth.

Have you seen any tailwinds yet or it's still too early to...

So we have seen, in some cases, we've seen 3 things. We've seen one data that is generated over the Internet, increasingly being stored, so it becomes useful for AI training. Two, we're seeing AI inference locations be established in peripheral locations around the Internet closer to end users with lower latency.

And third, we are at the very early stages of seeing end users being able to put information into those inference models. That data will actually be uploaded to those edge sites, processed against those large language models and the results will be distributed back to those users. That process today is fairly early on, and we have just started to see in certain limited business models AI be integrated into their core offering. I think as general AI becomes more pervasive, it will become embedded in every application that we use.

And then putting it all together, you've talked about growing EBITDA sequentially each quarter, putting the T-Mobile payments aside. Can you just update us as to how you're thinking about 4Q and maybe an early look at 2026 as to what you see as the big drivers of EBITDA into next year? .

So we try to give multiyear guidance. And our guidance is that we will grow top line revenues 6% to 8%, and we will expand EBITDA margins by about 200 basis points a year. We still have a net present value of payments due from T-Mobile to Cogent of about $224 million. We are receiving $25 million a quarter from them. And with those supplemental payments, our EBITDA margins today are about 31%, still below the 40% that we had pre-acquisition, but growing faster than 200 basis points a quarter.

We expect fourth quarter and then throughout next year to continue to show an improvement in top line growth an expansion in margin and growth in EBITDA that will be mid-double digits and the ability to reach our net leverage target of 4x. So at that point, we can resume the increase in our dividend.

We had 52 sequential quarters of growing our dividend and dividing it out more than 100% of cash flow by levering up incremental EBITDA. Cogent organically grew for 18 years as a public company with no M&A at 10.2% with an average of 220 basis points a year of margin expansion. We took a onetime step back, which was not unexpected. We are now actually at a point where top line is growing, and we'll continue to see that 200 basis points of margin expansion.

Just on the dividend, can you just help frame for us what changed in the thought process relative to earlier this year?

So I think there were 2 key drivers, Chris, one prior to reporting Q3 it was clear that our dividend had become decoupled from our share price. Our dividend yield was nearly 11%. And therefore, we felt that investors did not believe in the durability of the dividend; two, our aggregate net leverage had increased to a peak of 6.6x.

We felt it was necessary to be able to demonstrate to debtholders that we were serious about returning to the net leverage that we had operated in for 13 years prior to Sprint. From 2010 through 2023, we were generally between 3.5x and 4x levered and growing our dividend. Now that our leverage is at 6.6, we need to demonstrate a commitment to delevering. We also preserve flexibility for buybacks. We made 2 carefully worded statements.

One, that we were going to reduce our dividend by 98% and keep it at that reduced level until we reach 4x leverage. And then two, we would be willing to do buybacks on an opportunistic basis after a short pause or a temporary pause, we have lifted that pause. So we do have the ability to do buybacks opportunistically.

So it's fair to think that you can be out there in the market ahead of the 4x leverage target? .

We could be doing buybacks subject to the limitation of $105 million cap, which is all that is authorized by the Board. We do not intend to increase that cap until we reach the 4x leverage.

Got it. And then this past quarter, you disclosed you had the letter of intent on 2 of your large data center facilities. Just where does that process stand? And what does the time line look like for monetizing those other data center assets that you've repurposed?

So we have a nonbinding LOI with a credible party for 2 of the 24 data centers that we have earmarked as surplus. The surplus data centers represent 109 megawatts of power, 1 million square feet of data center arrays for space. We have other letters of intent that we're negotiating with parties. I think we will be successful in selling many of these facilities.

Again, to just remind investors, when we acquired Sprint, as part of the acquisition, we got 482 fee simple loan buildings that totaled 1.9 million square feet and had 230 megawatts of inbound power. These were primarily long-distance switch sites. We had initially anticipated taking 45 of the larger facilities and putting a 1-megawatt 10,000 square foot retail colo in them and leaving the remainder of the power follow. We then pivoted as it became clear to us that power was in short supply.

We decided to invest $100 million converting the negative 48 DC plants to AC and repurposing these facilities. And that work was completed in the end of June of '25 and that has allowed us to get to the letters of intent that I spoke about. The counterparty to that one that was announced, indicated they want to close in Q1. It is a nonbinding letter of intent. So I want to be clear, we don't have an absolute lever over them while they have an earnest money deposit, it is refundable.

And then you also sit on a large portfolio of unused IPV4 addresses. Why not sell those to delever faster?

So if we sell the addresses we get dollar for dollar benefit. We have no basis in them, so we would have some tax friction in that. If we lease the addresses out they carry virtually 100% incremental margin, and we get $4 of delevering, if our goal is 4:1 for each dollar of recurring revenue. We have seen a meaningful acceleration in our IPV4 leasing business. That business in 2022 was about a $12 million annual run rate.

As of last quarter, that business was about a $65 million run rate. So in 3.5 years, we've grown it from $12 million to $65 million. We expect that business to continue to grow. We do have a total inventory of 38 million addresses. Approximately $30 million of those are securitized today, and we've raised $380 million against them.

I think the ability to raise ABS is a better use of those address than selling in this market when the 2 largest buyers of address space have not been buying and are in fact competing with us in the leasing market at a price that is a significant premium to the rate at which we're leasing .

And maybe just one last one. You're coming off a period where you had been investing in either repurposing the Sprint wireline assets for the Wave business and then the data center facilities -- you've talked about this $100 million of core CapEx being a good run rate. Is there anything else on the horizon that we should be mindful of that could cause CapEx to tick up relative to that $100 million? .

So the answer is no. Although I do want to remind investors we do spend about $40 million in addition to that on principal payments on capital leases. The allocation of capital lease and straight CapEx is dependent on the nature of the asset. So you should think about $140 million just appearing in 2 different line items on the cash flow statement. $100 million of straight CapEx, $40 million of principal payments on capital leases.

I think that's all we have time for. Thanks, Dave, for being with us. .

Thanks, Chris, for hosting me.

Thank you, everyone, for joining us, and welcome to the Bank of America 2025 Leveraged Finance Conference. I'm Ana Goshko. I cover high-yield telecom and technology, and we're thrilled to have Cogent with us this morning and David Schaeffer, the company's CEO. Thank you, David, for being here as always. We appreciate you making the trek to Florida.

Thanks for hosting. Thanks for BofA for a great venue, and thank all the investors for taking time to hear a little bit about us.

Okay. Great. So I believe our audience here has a good background in Cogent. I thought that we kind of dive into some of the top questions.

Sure.

So the first one is really the wavelength business and the ability to scale the business and when we should expect to see a ramp. So just as a little bit of background. So you've cited strong wavelength circuit connection demand still hasn't really showed up in the revenue. And you've also said that you have a key customer that even though you've been installing circuits, they haven't been ready to accept the circuits.

So just to put it in context, I think your 3Q wavelength revenue increased $1 million sequentially to $10 million. That's now $40 million annualized. The prior goal was to exit the year about $100 million annualized. So if you can just provide us some context about what's been kind of dragging that down in terms of the ramp and when you expect to achieve that first goal of the $100 million or the $25 million kind of a quarter annual -- of quarterly wavelength revenue?

Yes. So let me kind of step back. The first thing we had to do was convert the Sprint network into a wavelength network and then connect that to an initial target of 800 data centers. We laid out that goal in September of '22 when we announced the deal. We began working on that in earnest in May of '23. And by December of '24, we had wave-enabled 802 data centers.

We've actually grown that footprint now to over 1,000 data centers where we can sell wavelengths. Those wavelengths can go from any data center to any data center. They can be provisioned in 30 days or less, and they can be delivered at 10, 100 or 400 gig speed. That gives us a unique footprint across the U.S., Canada and Mexico that is larger than any other provider.

During that initial period, when we were enabling the network, we initially targeted a subset of those 800 data centers. In fact, 65 of the largest data centers in North America and started selling waves in that footprint. We actually did sell a 1,000 waves before we had the entire footprint complete.

Remember, the Wave business is new to Cogent, and it did not exist inside of Sprint. It is a brand-new business. It's being deployed on a brand-new network. We had actually hoped to sell about three times as many waves in that core footprint as we did. Since the beginning of the year, in the first 9 months, we sold about another 800 waves, and we have gotten the quarterly revenue up to approximately $10 million. The revenue grew rapidly on a percentage basis, 93% year-over-year and 14% sequentially.

The wavelengths that we have sold came out of the backlog that we initially built during that development phase. Much of that backlog disappeared. We have rebuilt a new backlog in that we expect to continue to deploy an increasing number of waves each quarter. We have a goal of hitting $500 million in revenue for 25% market share by midyear '28, so call it 2.5 years from now. In Q3 of '25, we were at a $40 million run rate. So it's a steep ramp to climb.

We have been installing waves more quickly than customers could accept them. It's not just one customer, it's many customers. Our competitors typically sell wavelengths on a network that is a multipurpose network. We took a very different approach to the market. Our IP network, which is the largest in the world and carries 25% of the world's traffic, is completely independent of the wavelength network. It is entirely built on IRU fiber and stretches 92,000 intercity miles -- excuse me, 32,000 route miles of metropolitan in 58 countries around the world.

Our wavelength network, which is exclusively in North America is predominantly built on owned fiber, on different pairs of fibers. And in terms of revenue growth, we had initially laid out a target to exit Q4 with a run rate of December [indiscernible] of $20 million to $25 million. It is unlikely we will hit that target in that time frame because some of the waves that we have installed have not yet been customer accepted. We do expect that distance or time between installation and acceptance to shrink. We have tried to change the customers' perception of the market.

For most customers, they are accustomed to a market where 50% of their wave orders never get fulfilled by the supplier. And when they are fulfilled, it's generally a 3- to 4-month installation window. The customer bears costs initially for space and power in the data center as well as a cross connect, and they do not want to bear those costs before the service is ready. We have been able to demonstrate to about 200 customers and 500 data centers so far that we can meet that 30-day window. And I think as we continue to build credibility, we will continue to see a larger portion of the wave market shown to us and see our Wave revenues accelerate.

Okay. So I go back to the original question because of the shortfall and being able to achieve the original year-end target, which was the $25 million a quarter exiting the year. Are you setting a new target? Or at this point...

Target remains the same which is we will be at a run rate of $500 million by midyear '28 in wavelengths.

Okay. And any -- but interim guidepost?

No interim guide.

No interim guidepost. Okay. Okay. And then so on what is considered the legacy business, though I don't want legacy to be pejorative, but the heritage, I'd say, the heritage Cogent business?

The majority of Cogent.

Which is the IP, the IP business. So just in terms of the trends there and the growth outlook. So if I look at the corporate customers, the corporate customer connections in 3Q '25 was actually lower than it was before you acquired the Sprint network. And you still had a decline in this last quarter. So what is driving the decline? And when do you expect that to inflect?

So Cogent's core business had been selling Internet connectivity and IP-based VPNs to corporate customers and to wholesale customers. That business had grown organically at the rate of about 10% a year for 17 years. When the pandemic hit that growth rate in the Corporate segment turned negative. Our growth rate went from positive 11% among our corporate customers to negative 9%.

Today, the growth rate of corporate services is about 3% for organic Cogent. When we acquired Sprint, we acquired 2 customer bases. They were either corporate customers or enterprise. They were buying 1 of 2 services, MPLS-based VPNs or DIA. The Sprint business represented 40% of the combined company's revenue. It was declining pre-acquisition at an average rate of 10.6% per year. We actually accelerated that rate of decline and the business that we acquired from T-Mobile for the past 9 quarters has declined at an annualized rate of 24.2%.

The Cogent business, which is both transit to NetCentric customers and corporate actually has accelerated slightly and has resulted in total top line revenue declining at only 2.4%. To directly answer your corporate question, the corporate on-net portion of our business is growing. The corporate off-net portion of our business continues to decline, primarily due to the intentional reduction in the number of Sprint off-net customers.

So prior to the acquisition, Cogent's corporate base by revenue was 60% on-net, 40% off. By units, it was 80% on and 20% off. After the acquisition, even with the attrition in the Sprint customer base, at the end of last quarter, 51% of our corporate business by revenue was off-net and only 49% on. So it has been a combination of migrating some customers from off to on, but a much larger impact on that has been the rate of attrition in those off-net customers. Virtually all of the revenue decline in the last quarter came from the off-net Corporate segment.

Okay. Staying on the topic of the corporate -- the classic corporate customer, do you have a geographic concentration with regard to your building footprint? And impacts that have potentially been a factor or things like weak return to office, actually Dodge and then increasingly concerns about AI impacting either slowed hiring or actually kind of a new round of corporate layoffs. Like to what degree are you -- do you feel that your business is exposed to these factors? And is there like a geographic concentration that would make you more exposed?

So our model for corporate users has always been to build our network into the largest and most tenant diverse buildings in the market. We are in approximately 1,870 buildings across North America, largest number of buildings in New York City, followed by Chicago, Toronto, L.A., pretty much following the population base. The average building that Cogent connects to is 550,000 square feet. Pre-pandemic, those buildings had an average of 51 discrete businesses.

Today, the occupancy rate in those buildings has declined, and there are only 38 unique tenants within the building. The vacancy rate in our footprint increased from 4% to 18% at peak. Today, it is still at 17%. So that absolutely presents a headwind to our corporate business. And our corporate on-net business growth rate is about 1/3 of what it was pre-pandemic. It had been growing at about 11% per year. Now it's growing at about 3% per year.

Okay. And I think you touched on some of this in your earlier comments, but just to kind of crystallize it. So on the NetCentric side of the classic Cogent business, so the overall NetCentric customer connections did return to growth in 2Q and 3Q and the ARPU has gone up as well. But I think -- and then so your base right now, so 3Q, the NetCentric revenue was $100 million in the quarter. But -- and that was up about $3 million sequentially, $8 million year-over-year. But I do think in that mix is the IPV4 leasing as well as the wavelengths are actually the biggest driver really of that NetCentric growth. So what is the dynamic among sort of your classic customers in the NetCentric business right now?

So our historical business was selling bulk Internet connectivity and carrier-neutral data centers. As you pointed out, our wavelength business contributed $1 million of that $3 million of sequential revenue growth. The IPV4 leasing is 84% NetCentric, 15% corporate and 1% enterprise. So it also contributed about $400,000 of that incremental, but about $1.6 million sequentially was incremental IP transit sales.

The primary driver of that revenue growth has been international as traffic growth in the developed world has slowed in the less developed markets, we've seen an acceleration in traffic growth. So when Cogent was founded in 1999, 85% of all Internet bits carried in the world originated and terminated in the U.S. Today, that number is under 30%. We operate in 58 countries around the world and have disproportionately gained traffic in some of the less developed markets where pricing per megabit is higher.

Okay. Switching to the overall outlook. So the legacy Sprint noncore business and costs since you acquired the Sprint wireline assets in early '23, you've been pruning both the undesirable noncore Sprint revenue as well as the costs. So where are you on that journey with regard to what's left still to do?

So we have taken about $220 million of direct costs out of that business between May of '23 and September of '25. We have about $20 million more to take out over the next 1.5 years or so. Secondly, we are still spending about $45 million annually on various integration projects as we continue to consolidate and optimize systems and personnel. We will expect those costs to also taper off and be completed by the end of '26.

Okay. So putting this all together, the wavelengths, the Heritage Cogent business and the remaining cost saves. So right now, about 1/3 of the reported adjusted EBITDA is -- comes from the IP transit payments that you get from T-Mobile, which is effectively a subsidy. So you got $100 million annualized. Right now, the EBITDA is about $300 million annualized. So you've got 2 more years on that. So I think you've got $224 million of contractual payments still to come by -- mostly by the end of '27. I think some of that might -- a little tail of that might come in '28. But -- so you basically got 2 years to make up $100 million of EBITDA that you'll be losing in 2 years. So when we think about the wavelengths, the growth in the Heritage Cogent and then the cost saves, what is the bridge? Like what mix is going to get you to make up that $100 million?

So let's just remind investors, we have grown EBITDA for 9 sequential quarters. We've grown quarterly underlying EBITDA without the subsidies from $3 million to $48 million in those 9 quarters. We will continue to get benefits from continued cost savings and integration efforts, but we are also returning to top line growth. We have busily pruned the noncore products. We also have accelerated the growth of on-net products. Those on-net products are Cogent's on-net corporate business as it's returned to that 3% growth rate.

Cogent's transit business, which is growing at about 8%, the $1.6 million sequentially that you just referenced off of a base of a $100 million. the growth in our IPV4 leasing, which is on-net, which is growing at nearly 20% per year. And our wavelength business, which today is small, but grew at 93% year-over-year. We expect total top line growth to get to between 6% and 8%. With that 6% to 8% top line growth, we will be able to see our EBITDA margins, absent the T-Mobile subsidies expand by better than 200 basis points a year.

Prior to the acquisition, Cogent saw its EBITDA margins organically grow from 0 when we went public in 2005 to 40.5%. We then took a massive step down due to the acquisition of a money-losing business. Our margins were reduced to below 3% EBITDA margins. We have recovered to 20% today without the subsidy payments. As those subsidy payments roll off in Q1 of '28, we do have $224 million, but that's actually discounted back. There's actually a little more than about $250 million in cash payments still due to us.

We will continue to grow the underlying EBITDA through the growth in those on-net services with very little to slightly negative growth in off-net services. We anticipate eventually growing at about 200 basis points a year from the 20% margins today to back up to the 40% that Cogent had pre-acquisition off of a larger base in the next 8 to 10 years. But when those payments step off, we will have about a year in which reported EBITDA will be flat as the underlying growth will replace the $100 million that falls off on an LTM calculated basis.

Okay. So switching to just a credit conference here. So the company recently cut or paused its dividend, which is pretty substantial in order to focus on deleveraging. And you have $750 million of unsecured notes due in '27. So what is the game plan for dealing with those notes you have cash on the balance sheet. You're going to have cash flow that otherwise would have gone to the dividend. And then I know you obviously had some data centers. You recently said you had a nonbinding letter of intent for a sale of 2 of them for $144 million.

What is the game plan in terms of addressing those notes? And might you do it sooner rather than later if you have the cash and the cash flow instead of kind of incurring the negative carry on just continuing to pay that interest expense?

Yes. So let me just remind investors, we had a dividend policy for 15 years that dividended out more than 100% of cash flow by growing EBITDA at about 18% a year. We were taking that growth in EBITDA and levering it and remaining levered between 3.5x and 4x EBITDA during that entire period. Because of the capital expenditures associated with the integration of the T-Mobile assets and the negative cash flow that we acquired, we have seen our leverage peak at 6.6x net leverage. As a result of that, we made the decision to dramatically reduce our dividend from $1 a share a quarter to $0.02 a quarter, and we committed to keeping that dividend reduced until such time as we reach 4x net leverage.

So we are very serious about continuing to delever, making that a priority. We did pause our buyback program and then said we would retain the flexibility to execute up to $105 million of buybacks that have been authorized, but there is no immediate plans to necessarily do that. It's just a flexibility that we preserved.

In terms of the current June 27, $750 million unsecured. We have multiple paths that we can take to repay that debt. One, we have about $400 million of incremental capacity at our secured borrowing entity and could do an additional secured offering. Two, we have about $150 million available in our current ABS that we could tack on to if we elected to take the B and C tranches that we initially did not take when we did those offerings. Third, we could refinance in the unsecured market. Fourth, we could look at an ABS structure for our fiber assets, which we have not done and has been invoked with many other fiber providers.

And then finally, we have capacity. It would have an equity component at the holdco level with a convert now that we don't have a material dividend dragging down that conversion premium. I think we're going to look at all of those options. I think we need to demonstrate to debt holders that we are serious about getting to that 4x leverage target that we laid out. I think the asset sales, coupled with the reduction in the dividend and the growth in EBITDA will prove that. We do have a little over 1.5 years to go.

In the past, we have gone as close to 3 months before maturity before we've refinanced. I think it's unlikely we'll go that long, but I also don't think there should be an expectation that we're immediately going to refinance the '27s sitting here in December of '25.

Okay. So you're going to keep us guessing.

No, no, keep our options on.

Okay. I think we're effectively out of time, but we could easily do another half hour. So thank you, Dave. All the best to you in 2027.

Thank you, Ana. Thank you all.

Very, very interesting. Thank you.

Take care.

Good morning, and welcome to the Cogent Communications Holdings Third Quarter 2025 Earnings Conference Call. As a reminder, this conference call is being recorded, and it will be available for replay at www.cogentco.com. A transcript of this conference call will be posted on the Cogent's website when it becomes available. Cogent's summary of financial and operational results attached to its press release can be downloaded from the Cogent website.

I would now like to turn the call over to Mr. Dave Schaeffer, Chairman and Chief Executive Officer of Cogent Communications Holdings.

Thank you, and good morning, everyone. Welcome to our third quarter 2025 earnings call. I'm Dave Schaeffer, Cogent's Chief Executive Officer. And with me on today's call is Tad Weed, our Chief Financial Officer.

I'd like to recognize a number of significant events in the quarter and discuss a few matters and then turn things over to Tad. First, our program of return of capital and our aggregate leverage. Following extensive discussions with our Board of Directors and engagement with shareholders and bondholders, we have refined our capital allocation priorities to strengthen our financial flexibility and accelerate our delevering strategy. The decision to reduce our quarterly dividend to $0.02 per share per quarter was made after careful evaluation and will allow us to redirect capital towards reducing leverage, while remaining a disciplined approach to shareholder returns.

Supported by our growth in EBITDA, recurring cash inflows under our IP Transit agreement and continued operational efficiencies, we will reduce our net leverage ratios. These actions position us for long-term growth and enhance the financial resiliency of our business. We intend to maintain our updated dividend policy until we reach a net leverage target of 4x EBITDA on an LTM basis. On our last earnings call, we stated that we believe our leverage had peaked on an LTM basis. We also indicated that amounts due to us under the T-Mobile payments on our transit agreement and purchase agreement should be considered in calculating our leverage ratios. We believe these amounts essentially represent both short-term and long-term cash amounts based on the credit quality of T-Mobile.

Our total gross debt as adjusted for these amounts of T-Mobile for the last 12 months EBITDA as adjusted ratio was 7.74 last quarter and was reduced to 7.45 this quarter. Our net leverage ratio was 6.61 last quarter and 6.65 this quarter. T-Mobile pays us $25 million each quarter through the fourth quarter of 2027 under the IP Transit services agreement. These payments will be -- will reduce the total amount due to us each quarter from T-Mobile. We are also temporarily suspending our stock buyback program.

Now, for a couple of comments on our data center divestiture and monetization. In early October, we entered into a nonbinding letter of intent with a credible counterparty to sell 2 of our larger data centers out of the 24 data centers that we had repurposed. This agreement calls for a cash payment of $144 million. The counterparty has demonstrated to us the ability to complete this transaction and is completing its due diligence. We are in the process of finalizing the asset purchase agreement. We intend to monetize all 24 of the data centers, either through outright sales as with this $144 million transaction for 2 facilities or by leasing the acquired space on a wholesale basis. We are in active discussions, negotiating LOIs with other parties that we feel are credible to purchase or lease these facilities.

Now, for a couple of comments on our wavelength trajectory. At quarter-end, we're offering wavelength services in 996 data centers with the capability of provisioning 10-gig, 100-gig and 400-gig services within a 30-day installation window. Our wavelength services revenue in the quarter was $10.2 million, an increase by approximately 93% on a year-over-year basis. And on a quarterly sequential basis, our wavelength revenue increased 12%. At the end of the quarter, we had sold and provisioned waves in 454 locations as opposed to the 418 data centers that we had installed waves in at the end of Q2. We currently have a backlog and funnel of wave opportunities of 5,221 opportunities. We intend to continue to capture market share and believe our goal of 25% of the highly concentrated long-haul wavelength market in North America in 3 years is achievable.

Our EBITDA increased sequentially to $48.8 million, and our EBITDA margin increased sequentially by 50 basis points to 20.2% from continued cost reductions and product optimization. Our EBITDA as adjusted increased to $73.8 million, and our EBITDA as adjusted margin increased sequentially by 70 basis points to 30.5%.

Our IPv4 leasing activity materially accelerated. Our IPv4 leasing revenue increased by 14.1% to $17.5 million on a sequential basis. And on a year-over-year basis, revenues from IPv4 leasing increased 55.5%. Our average revenue per IPv4 leased in the quarter was $0.31 per address due to some larger wholesale leasing activity. We were able to accelerate this growth by entering into these more flexible agreements. We were leasing 14.6 million addresses at the end of the quarter, a sequential increase in the number of leased addresses of 10.7%. We have in inventory a total universe of approximately 38 million addresses.

We have essentially completed the integration of the Sprint network and the repurposing of the facilities that we deem appropriate for data center activity. As a result of this, there was a significant reduction in our capital expenditures in the quarter. At quarter-end, we were providing services in 1,686 carrier-neutral data centers and 186 Cogent data centers. The Cogent data centers have an aggregate of 214 megawatts of installed and available power.

We anticipate our long-term annual revenue growth rates will be between 6% and 8% and an increase in our EBITDA as adjusted margins of approximately 200 basis points per year. Our updated revenue and EBITDA guidance are intended to be multiyear goals and are not intended to be used as specific quarterly or annual guidance.

Comment on our revenue. We are nearing the end of grooming of our low-margin Sprint-acquired contracts. As we have stated on our last earnings call, we expect to return to total revenue growth by mid-third quarter 2025. However, our revenue for the quarter declined at $4.3 million or 1.7%. For the quarter, we experienced a $1.3 million decline in noncore revenues and an additional decline of $800,000 in USF revenues. Our high contribution on-net services and wavelength services, both increased in the quarter. Our on-net revenues increased by $2.9 million sequentially or 2.2% from last quarter. Our wavelength services revenue increased by $1.1 million or 12.4% from last quarter. And our IPv4 leasing revenue increased by $2.12 million or 14.1% from the previous quarter. We remain highly focused on selling products that deliver higher margins and allow our EBITDA margins to continue to expand.

Now, I'd like to turn it over to Tad to read our safe harbor language, provide some additional operational metrics, and then we'll close with a few closing remarks and then open the floor for questions and answers.

Thank you, Dave, and good morning, everyone. This earnings conference call includes forward-looking statements. These forward-looking statements are based on our current intent, belief and expectations. These forward-looking statements and all other statements that may be made on this call that are not historical facts are subject to a number of risks and uncertainties, and actual results may differ materially. Please refer to our SEC filings for more information on the factors that could cause actual results to differ. Cogent undertakes no obligation to update or revise our forward-looking statements.

If we use non-GAAP financial measures during this call, you will find these reconciled to the corresponding GAAP measurement in our earnings releases that are posted on our website at cogentco.com.

Some comments on overall results. Our revenue for the quarter was $241.9 million. Our EBITDA as adjusted was $73.8 million for the quarter, an increase of $0.3 million, and our EBITDA as adjusted margin increased sequentially by 70 basis points to 30.5%. Our EBITDA as adjusted accounts for payments under our IP Transit agreement with T-Mobile. Under this agreement, we received 3 monthly payments totaling $25 million in this quarter and the same as last quarter. We will continue to receive an additional 26 monthly payments of $8.3 million until November of 2027. There are further cash payments related to lease obligations that we assumed at closing. That totals at least $28 million. This $28 million will be paid to us in 4 equal payments from December 27 to March 2028.

We analyze our revenues based upon network connection type, which is on-net, off-net, wavelength and noncore, and we analyze our revenues based upon customer type. We classify our customers into 3 types: NetCentric, corporate and enterprise.

Our corporate business represented 43.5% of our revenues for the quarter. Our corporate revenues decreased by 9.5% year-over-year and sequentially by 3.5%. These decreases in corporate revenue are primarily due to the continued grooming of low-margin off-net customer connections and the continued elimination of acquired noncore products acquired with Sprint Wireline.

Our NetCentric business continues to benefit from the growth in video traffic, activity related to artificial intelligence, streaming, IPv4 leasing and wavelength sales. Our NetCentric business represented 41.4% of our revenues for the quarter. Our NetCentric revenues increased by 9.2% year-over-year and sequentially by 3.1%.

Our enterprise business represented 15.1% of our revenues for the quarter. Our quarterly enterprise revenue decreased by 25.7% year-over-year and sequentially by 8.6% due to a reduction in acquired noncore and off-net low-margin enterprise revenues acquired with Sprint Wireline.

On-net revenues: we serve our on-net customers in 3,537 total on-net buildings. Our on-net revenue was $135.3 million for the quarter, a small year-over-year decrease of 0.9%, but a sequential increase of $2.9 million or 2.2%.

Off-net revenue: our low-margin off-net revenue was $95.1 million for the quarter. That was a year-over-year decrease of 14.5% and a sequential decrease of 6.9%. It was $7.1 million of the decrease sequentially. We serve these 25,518 off-net customers and 18,400 off-net buildings. Our off-net revenue results are impacted by our migration of off-net customers to on-net and the continued grooming and termination of acquired low-margin off-net contracts acquired with Sprint Wireline.

Some comments on pricing. Our average price per megabit for our installed base decreased sequentially by 10% to $0.16 and decreased by 31% year-over-year, both amounts in line with historical trends. Our average price per megabit for our new customer contracts for the quarter was $0.07, a sequential price per megabit decrease of 8% and 15% year-over-year.

Our ARPUs for the quarter: our ARPUs for the quarter were as follows. Our on-net ARPU was $515. Our off-net ARPU was $1,225. Our wavelength ARPU was $2,108. Our IPv4 ARPU was $0.31 per address for the quarter.

Churn rates: our on-net and off-net churn rates both improved marginally from last quarter. Our on-net unit churn rate was 1.3% compared to 1.4% last quarter, and our off-net churn rate was 2.1%, a slight improvement from 2.3% last quarter.

Traffic: our IP network traffic growth accelerated for the quarter. Our network traffic increased by 5% sequentially and year-over-year by 9%.

Rep productivity metrics: our rep productivity was 4.6 this quarter. It was 4.8 last quarter and 4.0 in the third quarter of last year.

Foreign exchange: our revenue earned outside of the U.S. is about 20% of our revenues for the quarter. The average euro to USD rate so far this quarter was $1.16 and the Canadian dollar rate is $0.72. Using these average rates, we estimate that the FX conversion impact on our sequential quarterly revenues would be a negative $200,000 and the impact of the year-over-year quarterly revenues would be a positive $2.6 million.

We believe that our revenue and customer base is not highly concentrated. Our top 25 customers represented about 16% of our revenues for the quarter.

CapEx and payments on capital leases, principal payments. Our CapEx declined by 35.5% sequentially and was $36.3 million this quarter, down $20 million from $56.2 million last quarter and down $23 million or 38.8% decrease from the third quarter of last year. Our principal payments on capital lease were $8.8 million this quarter, similar to $8.5 million last quarter.

Some comments on debt and our debt ratios. Our total gross debt at par, including $601.8 million of finance lease obligations, was $2.3 billion at quarter-end. And our net debt -- total net debt [ of our cash ] and our $224.2 million due from T-Mobile was $1.9 billion. Our leverage ratio as calculated under our more restrictive covenants on our unsecured $750 million 2027 notes indenture was 5.66, and our secured leverage ratio was 3.49. Our fixed coverage ratio was 2.62. The definition of consolidated cash flow under our $600 million secured 2032 notes indenture includes payments -- cash payments under the IP Transit Services agreement with T-Mobile in that definition and determination of consolidated cash flow. And those payments totaled $100 million for the last 12 months. As a result, our leverage ratio as calculated under the $600 million 2032 notes indenture was 4.39. Our secured leverage ratio was 2.7, and our fixed coverage ratio was 3.38.

Lastly, some comments on bad debt and DSO. Our DSO improved. It was 30 days at quarter-end compared to 31 days last quarter. Our bad debt expense was less than 1% of our revenues for the quarter. It did increase from last quarter, about $1.2 million, which is not material, but it was only 0.5% of our revenues for the quarter. Last quarter, our bad debt expense was artificially low because we had some bad debt recoveries, which helped our SG&A expenses. But overall, our target and our historical rate is 1% of revenues, and we are -- our results are better than that historical rate.

And with that, I will turn the call back over to Dave.

Thanks, Tad. I'd like to highlight just a couple of strengths around our network, customer base and sales force. We continue to be beneficiaries of continued migration to over-the-top video, artificial intelligence activities and various other streaming trends. At quarter-end, we were able to sell wavelength services in nearly 1,000 carrier-neutral data centers with expedited delivery capabilities. At quarter-end, we sold IP services in a total of 1,872 data centers across the 57 countries and 302 markets in which we operate. At quarter's end, we were directly connected to 8,043 networks. 22 of these networks represent peers, and 8,021 of these networks are, in fact, paying Cogent transit customers.

We continue to remain focused on improving the productivity of our sales force and managing out those reps that are underperforming. Our sales force turnover rate was 6.6% a month in the quarter, down from a peak of 8.7% during the height of the pandemic. It is, however, above our historical average turnover rate of 5.7% of the sales force per month. At quarter's end, we had 617 quota-bearing reps. Our sales force includes 294 sales professionals focused on the NetCentric market, 309 sales professionals focused on the corporate market and 14 sales professionals focused on the enterprise market.

We have modified our return of capital program, allowing us to accelerate the pace of our delevering. We are actively working to monetize our acquired Sprint assets such as the data centers that we do not feel are core to our business and will allow us to further accelerate that delevering and allow us to resume our return of capital to our equity holders.

We have effectively completed the integration of the former Sprint network and buildings into the Cogent infrastructure, and now, we operate a single unified global network. We have completed the conversion of the Sprint facilities that we deemed appropriate to be converted to data centers. And as a result of this, we anticipate a continued reduced level of capital spending going forward.

We are optimistic and enthusiastic about our wavelength services business. While we have installed more wavelengths than customers have actually accepted, we are seeing changes in customer behavior and believe we will be able to accelerate the revenue recognition from these wavelength services going forward. Our wavelength services are differentiated by the quality and reliability, the uniqueness of the routes, ubiquity of our footprint and the speed at which we can install these services. Since our inception, we have built our business on offering superior services, expedited provisioning and disruptive pricing. That is why Cogent is a market leader in the services we sell.

With that, I'd like to open the floor for questions.

[Operator Instructions] Your first question comes from the line of Greg Williams with TD Cowen.

Dave, first question is on the dividend cut. Do you see the company returning to a $4 dividend level as you reach that 4x leverage? Or would it sort of ramp up and do you see it at that level at some point in time? And I ask that because with your diminished ownership, just trying to understand if a sizable dividend is still part of the DNA of the stock.

Second question is on the data center sale. Nice to see a couple sold here. Can you help us with the valuation? How much was it per megawatt? And when you think about the remaining 22 data centers or potential data centers, could they set similar valuations? Or did you just sell sort of the higher tier ones? I know you said these 2 were larger than the others. So I was wondering if that valuation will be sort of sustainable with the remaining facilities.

Yes, sure. Let me try to answer both of your excellent questions, Greg. First of all, on the reduction in the dividend until we reach a net leverage target of 4x. This was not taken lightly. It was actively debated by the Board, and we weighed feedback from equity holders, as well as feedback from bondholders, and we felt in aggregate, this was the best strategy. Two, once we have reached that target, we are committed to continuing to return capital to shareholders at a similar rate to what we were doing previously. What I cannot commit to is an exact restart where we left off on a per share per quarter rate. We also may be more aggressive in doing buybacks than dividends. But our commitment is to return all surplus capital to shareholders beyond the capital that is needed to run the business. As you can see from our accelerating rate of EBITDA growth that we will be able to delever more rapidly now and then continue that EBITDA growth to be able to return even larger amounts of capital per quarter to shareholders ultimately.

Now, with regard to the data center sales, we have a number of other LOIs in negotiations. We have, I think, groomed the pool of parties we're talking to, to only those that we feel comfortable can actually perform. The 2 facilities that have been agreed to are representative of the base. We are very pleased with the price per megawatt, and it is definitely within the range that we would have anticipated for the entire portfolio. However, I am not comfortable in disclosing that because that would then set a benchmark for our other negotiations. And in some cases, some of those negotiations are going to yield higher prices per megawatt. Each facility is slightly different based on its size, current power availability and potential for future power augmentations. So I'm reluctant to put an exact price per megawatt out into the market because then I'm locking, I think, the company into capping what we will receive on the other negotiations. And at this time, I expect at least some of the other facilities could potentially yield even higher prices per megawatt based on the unique characteristics of those facilities.

Your next question comes from the line of Frank Louthan with Raymond James.

Great. Can you talk to us a little bit more about the run rate on the waves? You'd stated hitting a run rate of $20 million to $25 million, I think, by year-end. Just talk to us about that. And then, on the products you're selling, is it mostly 100-gig waves? And any thoughts on your ability to capitalize on demand for 400-gig and 800-gig waves with your current network?

Yes. Sure, Frank. I'm going to actually take those in somewhat reverse order. We are capable today of selling 10-gig, 100-gig and 400-gig across the entire footprint that we have outlined. While the equipment in our network can actually support 800-gig and even 1.6 terabit interfaces, those interfaces for customers are not readily commercially available, and there really is no commercial market today for 800-gig. But our network at all sites will be capable all the way up to 1.6 terabits per wave as that market develops.

In terms of the mix, today, roughly 79% of our wave sales have been at 100-gig. That is very different than the aggregate market, which is today dominated by 10-gig wavelengths. And there is a product rotation that is ongoing across the industry where customers who had previously had 10-gig waves are now migrating to 100-gig. And then, there is further migration from 100-gig to 400-gig waves. Today, just under 10% of our sales have been at 400-gig, but we expect that to continue to accelerate. We have the capability to provide any to any data center connectivity. I know you wrote an extensive research piece on the total market for wavelengths. While we can serve the long-haul, the regional and the metro market, our greatest competitive strength is in the long-haul market because of the uniqueness of our routes, but we will be able to sell an end-to-end wavelength product that will include both metro and regional, as well as the long-haul, in a single unified product, which gives us a large addressable market.

Now, with regard to the acceleration in wavelength sales, we are continuing to build a funnel. We installed more wavelengths this quarter than the previous quarter and expect that trend to continue. As I commented in my prepared remarks, we have still installed more wavelengths than customers have accepted. We are actually encouraged by the fact that there's been a competitive response by other providers in the market to shorten their provisioning windows. That may sound counterintuitive, but it's actually helping us condition customers to take wavelengths as quickly as we can provision them.

Now, with regard to our exit run rate, it is highly dependent on customers' acceptance. To just clarify, we had talked about a monthly exit run rate at the end of Q4 that would get us to a quarterly rate of $20 million to $25 million. It is extremely dependent on the customers' acceptance of the waves that we have installed. At this point, we don't have enough visibility sitting here, the first week in November, to absolutely say that the backlog that we have installed will actually be accepted by year-end, but we are hopeful. Hopefully, that answered all your questions, Frank. Okay.

Great. And are you still confident in your long-term goal of wavelength revenue on an annual basis?

Yes. We absolutely believe that we will capture 25% of the addressable market. And while the entire North American addressable market is $3.5 billion today and growing because of AI demand, we feel that the $2 billion, which represents the long-haul portion of the market, is where we have the greatest competitive advantage, and we should be able to capture 25% of that or a $500 million run rate.

Your next question comes from the line of Walter [ Piecyk ] with LightShed Ventures.

It's Piecyk and LightShed Partners in this particular role, as you know, Dave. Just on the wavelength, so the $20 million run rate, I think you talked about last quarter exiting the year, or maybe I missed it when you were just responding. Is that -- are you expecting to hit that? And I think there was some maybe a little confusion last quarter whether that doesn't necessarily imply $20 million for the quarter, just the run rate that you're exiting as of December 31.

I'll take those again in reverse order, Walt. To be clear, and I thought I was clear enough both on the call and in subsequent conversations, it was a monthly run rate that would get us there, so not a revenue run rate for the full quarter, but an exit run rate. And then two, just reiterating what I said in answering Frank's question, to the first part of your question, while we will install enough wavelengths to hit that target, I am not today confident enough that the customers will accept them and we can start recognizing that revenue, although...

Even for the $20 million run rate, yes. I understand. Got it. Understood.

And we are encouraged by the fact that the gap between installation and acceptance is shrinking.

Got it. Dave, there's been some -- I mean, over the course of the quarter, since you had to sell your shares, there's been some conversation about how you get re-upped and how that process works. And obviously, you cut the dividend today, which probably is going to have an impact on the stock, which would present an attractive entry point, although, from a shareholder perspective, there might be questions about that, like if the Board re-ups you after driving the stock down. Is there any discussion in terms of your additional stock getting tied to perhaps the $500 million wavelength target that you have set out for the middle of 2028? Or should we expect the Board just to re-up you at this depressed level in the stock?

So those are ongoing discussions. I am committed to Cogent and currently do not have a contract that will extend beyond the end of the year, but I am in discussions with the Compensation Committee and then ultimately, the entire Board around that. The exact form of those incentives will be based on a number of operational metrics, as they've been in the past. And just to remind you that many management teams set the bar low, so they always hit the performance targets. In my case, I have been, in multiple instances, forfeiting shares because those targets have not been met. And the exact nature of how that program will be put in place is just still an ongoing discussion.

Okay. I think you understand maybe the perception issues if it's an end-of-the-year thing not tied to future numbers. But again, that's for the Board to decide. Dave, I want to go back to kind of our core here or what used to be the core. Obviously, wavelength is kind of a new growth opportunity and maybe IP addresses and data centers. In corporate, I saw on Bloomberg that in New York, like vacancies were like at an all-time low compared to COVID. I realize there's some Sprint trimming and there's some, whatever, not on-net stuff that's trimming, but 10% growth in corporate. Like, why is that still happening? And when can that invert or at least stop declining?

So our off-net corporate business declined $7.1 million on a quarter-over-quarter basis. The vast majority of that was acquired Sprint off-net corporate customers. The underlying Cogent corporate on-net business is growing in low-single digits, roughly about 3%. That is clearly not where we were pre-pandemic when that segment of our business was growing at nearly 11%, but it is a recovery from the low point. In the Sprint customer base, we acquired a mix of corporate and enterprise customers, virtually no NetCentric customers. And virtually all of that business was off-net. We have been aggressively managing out less profitable revenue. It's how we've been able to grow margins even though we have had top line declines that continued this quarter and will continue. There is probably some customer circuits that we lose that we wish we didn't lose because the customer has 5 locations, 3 of which are gross margin negative, and we either raise prices or ask them to turn those services off. They may also turn off the 2 locations that are acceptable margin and we would like to keep. The $7.1 million corporate off-net decline sequentially was greater than we had anticipated. In aggregate, since we have acquired Sprint, the rate of revenue decline has been over double what it was going into the acquisition because of this intentional grooming. We are near the end of what we can groom. The noncore revenue decline sequentially of $1.3 million, if we declined at that rate again, we'd have negative revenue.

There was only $1.4 million left.

Yes. It's all gone. And I think this is helping us drive margin expansion.

If I could just sneak one last one in, Dave. On the data centers LOI on the announcement, what type of due diligence do they have left? And I assume there's some taxation on that or maybe you have some -- I guess, you don't have NOLs, right? You've been giving the cash back through the dividend. But yes, just on the due diligence, what do they have to do? What are the -- what's our risk that it doesn't get finalized?

So there is a detailed list of diligence items. They have third-party consultants verifying the data that we have provided them, inclusive of third parties going and doing site inspections. Probably the most important of those diligence items is actually the verification from the utility of what the utility has verified to us in terms of serving power availability. Because these facilities have been effectively dormant for close to a decade, we initially, before we spent the capital, reached out to each of the serving utilities and confirmed that if we spent the capital that the power would be available. We got those affirmations, and the counterparties that are acquiring these facilities are doing that same utility verification. That's probably the most important point.

And then, in terms of taxes, you are correct, Walt, we have very few usable NOLs left. While we have a significant number of NOLs, they are mostly outside of the U.S., and there is virtually no tax basis in these facilities other than the capital that we have spent to modernize them. And because of bonus depreciation, we have very little basis in these facilities.

Got it. So what's the tax rate on that? Just like 20%, 30%, 40%? What is it?

Well, company tax rate is 25%, effective income tax rate.

Even on sales like that? Yes, okay.

Yes. That's federal and state. Yes, that's right.

But with bonus depreciation going forward, we do not expect, even if this closes with the $144 million, that we would have a material income tax liability.

Because we have the bonus depreciation.

Right. With the [ tax bill ].

Your next question comes from the line of Chris Schoell with UBS Financial.

Maybe just a follow-up on the wave competition. You've talked about a mix of market concentration, route diversity and faster provisioning driving your competitive edge. With Lumen enhancing its provisioning timelines and expanding its wave network, is this taking away from your edge? And has it had any ability -- impact on your ability to scale so far?

So, thanks for the question, Chris. So the answer is, not yet, but we're only 1.5% of the addressable market today, and Lumen is the dominant player. Our largest competitive advantages actually come from the route diversity that we offer and the reliability of our network vis-a-vis others. It is hard to count that as a competitive advantage until you actually have installed customers. So in kind of an early deployment, we had to focus on things such as provisioning speed and ubiquity of footprint. We still have over 3x as many data centers as Lumen that we can connect and provision in what they are alleging their improved provisioning times are in a subset of their data centers. But what they don't have is the uniqueness of footprint that we got from Sprint and the fact that the Sprint network was deployed with a much deeper buried cable that has been cut much less frequently, therefore, resulting in higher reliability, better throughput and ultimately, less future cuts going forward. We monitor the cut activity on a per kilometer basis versus the fiber that we have bought through IRUs. We have bought fiber now from 378 different suppliers around the world, roughly 124,000 route miles of IRU fiber. And across that footprint, the frequency of fiber cuts from IRU fiber, not calling out Lumen specifically, but across the entire base, is 7x that of what we experienced on the Sprint network. And I think that will ultimately be probably our biggest long-term competitive advantage other than pricing. But we feel very comfortable that we're going to continue to gain market share.

Got it. If I could just fit in one more. You still sit on a large portfolio of unleased IPv4 addresses. Can you just help us think through why not sell these excess addresses and accelerate the pace of delevering? And I appreciate, the data centers required investment and due diligence, but would selling this excess inventory be a simple and faster process?

So it would. However, 2 points. The first one is, we have modified our strategy on leasing that saw a material acceleration in our leasing revenue from these addresses. We had previously not leased out addresses allowing the counterparty to sublease them. We removed that restriction last quarter and entered into one large wholesale leasing agreement kind of mid-quarter. That was partially reflected in our revenue. And we have a second one of those agreements [indiscernible] today. We think that's a significant additional addressable market. And while it does yield a lower revenue per address per month, it does allow us to deplete the unleased inventory much more rapidly than through our direct sales efforts by going through these brokerage or wholesale type counterparties. We then can use that revenue to raise ABS capital at very attractive rates. And the alternative would be an outright sale, which to Walt's point, would have tax consequences because we have no basis in these. But also the market for these addresses has softened on a sale basis because the 2 largest buyers in the market have not been active buyers in the past 24 months. And as a result of that, while there is a broad market, I'm not sure the market is deep enough to absorb the type of volume that we would bring to market. So I think our leasing strategy at this point is the best way for us to generate cash and provide incremental financial flexibility.

Your next question comes from the line of Nick Del Deo with MoffettNathanson.

First, regarding wavelengths, I think in your remarks, you said that you installed more than have been accepted by customers, you're observing changes in customer behavior, and you're confident that you will accelerate the revenue recognition from waves. I guess, can you expand on those comments a bit, and as part of that, perhaps share the number of provisioned but not yet billed waves that you have to give folks comfort in the outlook you've shared?

Yes, sure. First of all, thanks for the questions. Let me start with the change in behavior. With our competitors offering accelerated provisioning, we are seeing customers now ordering waves with the expectation that they'll be delivered in a shorter window. Now, some of that could be the confidence that we're building with that customer based on our existing installed track record. Some of it could be the belief that the industry is changing. It's hard for us to disaggregate that. In terms of installed but not billed, it's several hundred waves that are in place today, sitting there waiting for customers to accept them. And with IP services, we do have a mechanism where we require the customer to start paying even whether they have affirmatively accepted or not. We are still probably several quarters away from implementing that same policy to our wavelength products. We are a new entrant. We have 1.5% market share. And what we don't want to do is alienate customers through forced billing that would then preclude them from giving us the opportunity to bid on a greater percentage of their wavelength demand.

The other thing that has been encouraging to us has been the fact that customers increasingly are choosing us because of the uniqueness of the route. It's something we expected. But until we had a large enough base, a couple of thousand waves installed, we needed to really hear from the customers why they're buying. And while the city pairs are available from any of our competitors, the actual routes that we're on in most instances, there is no other provider on those routes. And that has turned into, I think, in many ways, today, our most significant advantage. And if I think longer term, I think the greatest advantage that we're going to have is just demonstrating to customers the infrequency of cuts on our network versus the networks that they're more accustomed to. So the measure of quality is multidimensional. I think we're trying to win in every one of those dimensions.

Okay. That's helpful color, Dave. You kind of -- along those lines, the backlog was up. It wasn't up a ton. I guess, to what degree is that a function of demand or how focused you are with certain customers for sales versus the dynamic you described where customers are ordering closer to when they need the waves because they understand that they'll be provisioned relatively quickly?

It's a little hard, again, to disaggregate what's driving customer behavior. I also think we have really tried to discipline the sales team around trying to sign orders that will install more quickly. We fully recognize that we can't go on continuously installing more than we recognize revenue for. And while the sales rep does not get paid a commission until the service is installed, we've put some additional incentives in place to help them drive customers that will accept quickly. It will take several months for all these programs to kind of play out.

And just to clarify, the commission is paid when billing starts.

That's correct.

Not on installation. They get paid when we get paid.

All right. That sounds good. And can I -- sorry, can I slip in one last question on a different topic. Dave, you alluded to this in your prepared remarks about the inflection to positive revenue growth in mid-Q3. I guess, it wasn't obvious if that happened in September versus August. But I guess, more importantly, should we expect full quarter revenue growth Q4 versus Q3?

I believe we will see positive revenue growth. It did happen late in the quarter. And the one caveat to that -- and we did see more off-net corporate churn in the quarter than we anticipated. Some of it was collateral damage to other services that we terminated intentionally. I think that's behind us, and we do anticipate returning to aggregate positive growth.

And noncore revenue is now down to $1.4 million.

Yes. We can't lose another $1.3 million sequentially if there's only $1.4 million total left. So we've got some things, and we don't anticipate the incremental headwind from USF as well. So I think there were just a confluence of things this quarter, whether it be late inflection towards the end of the quarter and these additional headwinds. But we are on the trajectory to total top line growth.

Your next question comes from the line of Michael Rollins with Citi.

Dave, just maybe a few follow-ups. First, can you share -- if you go by your segments, corporate, NetCentric, et cetera, enterprise, can you share how much of the legacy revenue subject to churn is left? And I have a few other follow-ups, if you don't mind, afterwards. So as we just think about like what the remaining pressure is in each of the pieces, including even in that $1.4 million left of noncore?

Yes. So there was virtually no NetCentric revenue acquired from Sprint. The vast majority of the revenue was enterprise and a smaller percentage, but not immaterial, was corporate. The aggregate rate of decline of acquired Sprint revenue has been nearly 24% annually. It was 10.6% average negative decline for the 3 years prior. We have groomed virtually everything we want to groom. What we don't have visibility to perfectly is for the business that is left from those customers, which we're happy with and the customers are still using, how much of those services will still churn off because the customer had other services that they wanted to keep with us that we could not keep. Of the noncore revenue, it's down to $1.4 million. It's -- probably 80% of its still left is ex Sprint. I'm not going to say there was 0 Cogent noncore. It was very small pre-acquisition. But yes, it's probably a couple of hundred thousand. And then, in the corporate off-net segment, there still remains some Sprint revenue and 100% of the enterprise revenues, which represent about 15% of our revenues, are still from the enterprise customer base.

And maybe secondly, you talked about the wave business in terms of the different dynamics of 10-gig, 100-gig, 400-gig. What are the ARPU differentials between these different points as the mix shifts more towards 100-gig and 400-gig and maybe more over time?

Yes. So the ARPUs are determined by 3 inputs: the physical length of the pack, the size of the wave and the duration of the contract. Kind of the range across all 3 of those metrics goes from about $500 a wave to about $8,000 a wave. Our ARPU in the most recent quarter was $21.08. And I'll just pick kind of a typical, say, 1,500-mile wave. If you did that at a 10-gig rate, it would probably be $600 to $700. At a 100-gig rate, it would probably be $1,600, $1,700. And if it's a 400-gig wave, it would probably be somewhere around $4,000. So, that just gives you a sense. But it's a little hard to answer it exactly because you've got to look at all 3 inputs on a wave-by-wave basis to get to exact pricing.

That context is really helpful. And maybe just one more. Going back to the dividend cut and pausing the buyback simultaneously, can you give us a little bit more insight into the conversations that changed at the Board level? Where, if you look at the variances in financials in the quarter versus the severity of the action that you're taking on capital allocation, can you give us maybe a list maybe in order of importance to the Board of what really changed and over the time frame that these conversations really accelerated?

Yes. So 2 very different timelines. The first one is on the reduction in the dividend. We have a very specific target before we reaccelerate the dividend, and that is a leverage target. For the buybacks, we view those as more episodic. They have been temporarily paused, but they are not necessarily tied to the same endpoint. The second point and maybe the more important one that you asked is what changed. And I would say, the 2 inputs that the Board looked at from 2 different capital markets weighed heavily on their decision. They looked at the trading of our secured bonds and the fact that they had traded off to about [ 96 ], and they were concerned that those bondholders were concerned with aggregate leverage. The unsecured bonds continue to trade above [ 99, like 99.5 ]. And we view that as a less relevant input because of the short duration. But because the secured debt had a 7-year maturity, I think the sell-off in those bonds really had the Board focus on that constituency in our capital stack as part of their decision process.

I think the second thing that the Board looked at was the violent reaction of the stock last quarter to my forced sales. Now ultimately, the stock recovered for the most part, but there was a clear decoupling of the dividend yield from the stock price. And I think the Board looked at that and said, it was an indication that the equity holders did not believe the sustainability. So with those inputs, I think the Board realized that it would be in the best interest of all of our stakeholders to put a firm marker in the ground around what an acceptable net leverage target would be, and then two, a program that would help us get there as quickly as possible.

Your next question comes from the line of Michael Ng with Goldman Sachs.

I just wanted to ask about wavelengths and customer acceptance. I just wanted to kind of revisit this concept. Are these orders that you have in hand and there's a certain performance obligation before a customer accepts and you get revenue? Or is this more about you guys converting that [ fallow in ] connected capacity, dark fiber to lighting it up and then expecting to get an order after having that installed capacity? Any thoughts there would be helpful.

Yes. Thanks for the questions, Michael. So we do not pre-provision any capacity. Each wavelength is built on an order-by-order basis. What we are referring to are wavelengths that customers have ordered, have signed contracts, we have installed, we have provided the customer the test results of that installation and service, and the customer says, I am not yet ready to begin to use the wave. So very different than, say, your cable service at home. When you call the cable company, they turn up the service, you start paying whether you have a television to watch it or not. We do a similar thing with our IP services because we have conditioned customers over a long period that we meet our installation windows. We do allow customers 2 windows where they can move out acceptance on IP and then there is forced billing. On wavelengths, we have not implemented any kind of forced billing, but the wavelengths that are installed have specific contracts. They have been installed to the specification that the customer requested and then the customer has informed us they're not yet ready to utilize the service.

Great. And just as a quick follow-up, when at least I think about the historical customer base of point-to-point, regional networks, international carriers, content distributors, including hyperscalers, like, any specific group of customers that you see that are driving the new aggregate demand for waves? Is it DCI and AI? Or is it kind of more traditional?

Yes. So there are legacy use cases that remain, but the largest incremental use case that is driving an acceleration in unit volume and an acceleration in aggregate revenue for the entire industry is coming from AI. And the AI demand is coming both from hyperscalers who have both AI and non-AI use cases, as well as neoclouds who are exclusively AI-driven. But that is, by far and away, the dominant driver of incremental demand in the industry.

And there are no further questions at this time. Mr. Dave Schaeffer, I will turn the call back to you for closing remarks.

I would like to thank everyone for their time. We will be at a couple of conferences coming up, and I look forward to seeing everyone in person. Again, thanks a lot. Take care all. Bye-bye.

Thank you. This does conclude today's conference call. You may now disconnect.

All right. Well, good morning, everybody. Welcome to the Goldman Sachs Communacopia and Technology Conference. I have the privilege of introducing Dave Schaeffer, who is the founder and CEO of Cogent, which he founded in 1999. My name is Mike Ng and I cover U.S. telecom here at Goldman Sachs. We have about 35 minutes for today's presentation, inclusive of any audience Q&A. But I'd like to start in saying thank you, Dave, for making it out here to the conference and spending some time with us.

Well, thank you, Mike, for hosting me. Thank Goldman Sachs for a great venue. I'd like to thank all the investors for taking time out of their day to hear a little bit about Cogent.

To start things off, I was wondering if you could just talk about the big picture 3- to 5-year view. What are some of the biggest developments that you're focused on? And where should we be spending our time as analysts and investors on the Cogent story?

Yes. So for those of you who have covered telecom for any long period of time, it has been a very tough sector for investors. The Internet has been very value disruptive to the traditional telecom ecosystem. And that's going to continue going forward. Cogent has differentiated itself and been able to grow in that market by focusing on a limited number of products where we have a structural advantage. The bulk of our revenue today, nearly 90% of it comes from selling Internet-based services. For 25 years, Cogent was exclusively a provider of Internet and VPN services over the Internet. That is going to continue at Cogent to be a significant part of our business, is going to continue to grow. And we are going to continue to capture market share. We have significant structural advantages based on our network architecture, the reach of our network into 57 countries, about 1,870 carrier-neutral data centers and over 1 billion square feet of multi-tenant office buildings where we sell to end users.

A second part of our business is selling that same product set to large enterprise customers. This was part of the acquisition that we received when we purchased the Sprint GMG business from T-Mobile. That is a business that was in decline. It was declining at 10.6% a year for the 3 previous years to our acquisition. It was almost exclusively an off-net business and it was a business that had a large number of noncore products. We have worked diligently in taking costs out of that business to streamlining the product portfolio and getting that business from a negative 80% margin to breakeven. Over the next couple of years, we will be able to get that business to about a 20% positive EBITDA margin. And it's a business that is predominantly off-net based on where these large multinationals buy service, is a business that probably will continue to decline at 1% to 2% a year but we will be able to harvest some cash. And maybe the third leg of the stool is the most exciting. It has been the repurposing of the Sprint physical assets in order to sell wavelength services.

So the Sprint network was built between 1985 and 1992 at a capital cost of $20.5 billion. That network was comprised of 19,000 route miles of intercity fiber, 1,200 route miles of metropolitan fiber and 482 buildings. It was exclusively designed initially to carry long-distance voice. That network has been dormant since 2015. We acquired that network from T-Mobile as part of the Sprint GMG acquisition. That acquisition had 2 components, a $700 million subsidy payment for taking over the declining of money-losing enterprise business and it sold us the physical asset base for $1. We spent 18 months repurposing that asset solely to deliver wavelengths. We have a start-up business in a public company, a wavelength business. We have enabled 938 sites as the end of the second quarter to get a wavelength that can be delivered within 30 days in any direction to any site in 1 of 3 speeds, 10 gig, 100 gig and 400 gig. The addressable market that we are going after is a $2 billion addressable market. It historically had 3 use cases, now 4.

The first use case was international network extensions. Second use case was regional network consolidation and aggregation and the third use case was content distribution. A fourth use case, which is growing rapidly is connecting AI training data centers to AI data that needs to be used to build large language models. This is driving an acceleration in wavelength demand. We have 5 structural advantages that our competitors cannot match. One, we have a network that goes to more endpoints. We've got an architecture we can provision faster. Three, our physical routes in 90% of the cases are unique to us. So we provide a level of protection and diversity. Fourth, because we acquired this asset for $1, we have the ability to price very aggressively to capture market share. And then fifth, because of the way in which the fiber was constructed, it is more reliable than our competitors.

Our ultimate goal over the next 3 years is to go from 1% market share in the intercity wavelength market to 25% market share, growing that business from a $36 million run rate at the end of last quarter to a $500 million run rate business. So within Cogent, there's a core business that's growing at about 5% a year with about 100 basis points of margin expansion. There's a legacy acquired business that is declining in low single digits with no margin today and margins improving. And then finally, there's this rapidly growing new business line that is going to contribute very high-margin revenue.

It's a fantastic overview, Dave. Maybe just on the wavelengths piece. So a $500 million annual revenue run rate, $2 billion TAM, 25% market share. What do you need to do to execute from where we are today at $36 million to get to that $500 million by 2028, key customer wins, investments in the infrastructure? Yes.

So the investments have been made. In fact, our capital intensity is declining. We built out a network that only requires 2 field deployments in order to provision each wavelength. We plug in a optic at each endpoint and then can automatically provision that wave. The architecture that we have deployed is optimized for one and only one product, wavelengths, that is very different than what our competitors do. They typically sell wavelengths off of a multiservice platform where they're selling remnant or excess capacity, which requires much more customized deployment, much longer deployment windows and oftentimes results in wavelengths actually being ordered and then never installed. What we need to do to get from $36 million to $500 million in annualized run rate is simple. We need to earn the credibility of the customers. The customers need the wavelengths. The market is concentrated today among a couple of key players.

The customers are dissatisfied with those players for 3 reasons: poor performance, poor installation and difficulty in getting pricing that they feel is reasonable. We have the opportunity to win on each of those fronts. We have an existing sales force of 650 sales reps. That sales force is divided into corporate end users and then those that focus on service providers. Virtually all of the wavelength growth is going to come from the service provider portion of our sales force, the roughly 300 sales reps that focus on customers that will buy waves. 3/4 of all wavelength buyers are all ready buying transit services from Cogent. So we have a pre-existing relationship. We've demonstrated the quality and superiority of our transit services. And all we're asking is for the opportunity to do the same in the wavelength market. If necessary, we will discount heavily if that's what it's going to take to win market share. But to date we have not yet had to be as aggressive as we originally planned.

And when you talk about these customers, just in terms of the type and category and flavor of these customers, I was just wondering if you could just talk about what that looks like? Are these hyperscale customers what I would call like AI cloud, neocloud type of customers? Or are these more Tier 2 cloud, Netflix, Salesforce, like those types of companies?

All of the above, plus some. So let's start with the legacy customer bases. One, they are companies that are international carriers extending their network. Telefonica, Telecom Italia, British Telecom, Deutsche Telekom, these are all companies that have bought wavelengths from us. They are regional network carriers, companies like a Metronet and Allo, a FirstLight, a Hotwire, all customers that have bought wavelengths from Cogent. They are traditional hyperscaler operations. Meta, Google, Amazon, Microsoft, all customers of Cogent historically, who are distributing content, Netflix, distributing content, Akamai, Fastly, as third-party CDNs, distributing content. So all of those legacy bases have been purchasing from Cogent.

And today, we're only getting a very small portion of their spend. As we prove in the ability to deliver and the reliability of the service, the percentage of their expenditures that will be shown the opportunity to bid on will increase. We have a ubiquitous footprint that covers that entire addressable market. And then for the new use case and the new use case comes from hyperscalers who have a second line of business, They're in the AI business. Meta, Microsoft, Google, Oracle, have all been very vocal about that. There are pure-play companies, OpenAI, Anthropic, Lambda Labs, [ NeoClouds ], ones that are really building a network solely for AI training, All of these are Cogent's addressable market. These are customers that are buying from us and will continue to increase their spend with us. So our view of the addressable market was probably more sober when we announced the deal than it is today. And we were burned in the dot-com boom 25 years ago.

Cogent was founded in '99. It was actually -- the boom was sparked at Communacopia in 1998 and when the CEO of [indiscernible] gave a speech. I was actually there in the audience. I was in a little kitty room with a little one-on-ones and he was in a big ballroom and John gave a speech and he said, the number of T1s that we are selling for Internet access is doubling every month for the past 6 months. That was actually a factually true statement. But what everybody inferred from that was that the Internet was going to double every month for the rest of time. And it then elicited a nearly $2 trillion investment boom that ended in a crash. So when we came to the wavelength market, we looked at 4 independent market studies that all size the market in North America intercity at about $2 billion. We then went to our existing sales force and said, I want a bottoms-up market analysis before we agree to bid on Sprint. I want you each to go to every one of the potential wavelength customers in your funnel and assess how many wavelengths over time and how much they will spend.

Well, those 2 numbers triangulated to the same number. Now if I looked at all of the third-party analysis in 2022, when we announced the deal, they were expecting the market to grow 5% to 7%. I actually discounted that and said this is a flat market and we will be successful. And the reason we took that pessimistic view is that's exactly what has happened in the transit market. We have grown to be the largest transit provider in the world. We carry 1/4 of the world's traffic. But the TAM for that service is only $1.5 billion and it hasn't grown. It's grown in units but has not grown in revenue. So we took that same sober approach to the wavelength market. Where we were surprised to the upside is that as AI training really started to take off, the wavelength demand is growing. And I do today believe the total addressable market is probably growing at somewhere between 5% and 10% a year. That's a good fact. It's easier to get to our goal in a growing market than a static market.

The second bonus from that growth came on the data center side. So Cogent connects to 1,870 third-party data centers in 57 countries around the world, more than any other provider. That's where we sell 97% of our transit services in that footprint. We also have 180 of our own data centers. It is a very small business in Cogent. It's a sub-$30 million business. And with the acute shortage of power, driven by AI training, that has created demand for some of these data centers. And our hope is we can sell off some of these facilities. We did ramp up investments but we were encouraged when customers who would use these facilities came to them and told us what features they would need added to what were former switch sites to convert them to data centers. That work is now done. We are in the process of trying to sell off those facilities. So we have, I think, had a couple of fairly significant positive surprises to the upside.

Great. Just on the new use cases, just as a follow-up. Could you just put a finer point on what some of those things are? The -- connecting storage to training facilities, that sounds like data center interconnect. Is there a part of this wavelength thesis that's really relying on enterprise inferencing to really take off in a meaningful way so that the content distributors and some of the traditional hyperscalers see increased demand for wavelength over this period of time through 2028?

Yes. So the building blocks of AI have been known since about 2010. But it took 15 years for AI really to burst into the mainstream and be something that is now integrated in most applications. And the raw material for AI is the data that was collected over the Internet. So the zettabyte or so of data that is stored around the world is the raw material for large language model development. And because the processing to do that is very power intensive, usually, the processing exists in a different location than where the data resides. If you are building a data center for the purpose of doing AI training, roughly half of the capital invested goes to the GPUs to do that training. The reason why AI is practical today is we have low enough cost compute. We have enough data to actually compute against to build the tokens necessary to create the large language models and the neural networks that will then be used for the inference phase. And we have the toolkits to be able to do that.

In fact, large language models are evolving very rapidly, typically on a 4- or 5-month refresh cycle, much faster than traditional software, which is making this much more efficient. But the inefficiency of locations is causing the need for wavelengths. While the Internet is how all this data was collected and will continue to be the mechanism of collection, it will also be the mechanism of distribution of outputs. The inference phase will occur over the Internet from distributed caches around the world. Latency is important for the inference phase.

Before the training phase, the training has to occur where the power and space is available. Even though it's 2.5x cheaper per bit mile to use the public Internet to move the bits than to use a wavelength, AI prefers to use the wavelengths because it does not have to buffer the traffic. The Internet is not latency determinant. What that means is you don't know exactly how long it's going to get to take from point A to point B. And a couple percent variance in [indiscernible] can force you to add 10% buffering to your compute power, since that's half of your capital expenditure that is highly inefficient.

So if you are trying to optimize the number of tokens produced, which is really the primary goal of large language model construction you're going to want to do that using wavelengths. An alternative can be dark fiber, which is even more expensive. There are 2 downsides to wavelengths or dark fiber. One, they're more expensive; two, they're not protected. They're vulnerable to any number of real-world inferences, cuts. A jumper gets disconnected in a data center and you're down. So almost all wavelengths are typically sold in pairs.

That's extremely important because the fact that our routes are diverse from the other 2 vendors, yet we have the same city pairs, gives us a huge structural advantage. The Sprint network was built along railways. Most of the fiber of the other wavelengths and dark fiber vendors is along public highway right of way. Railroads and highways go to the same cities but on typically different physical paths. That spatial separation of a mile or 2 has tremendous value for resiliency and diversity. So the use cases are to get data out of a data repository to a training facility and then return that data once the tokens have been created.

Great. You talked a little bit about the preparation for the data center asset sales that you acquired as part of the Sprint deal. Could you just touch on how much that could potentially be worth? And then similarly, I think there are some excess IP addresses that you're also considering monetizing. What could that do for your capital structure and success?

Yes. So we have 3 assets that we've identified is not core to our operating business. The first of them are former Sprint tandem switch sites that are being converted to data centers. They -- there are 24 large sites that have a total of 109 megawatts of existing inbound protected power. We have gone through those sites and converted them from negative 48 DC to AC 120. We've refurbished air conditioners, fire suppression systems, distribution frames, generators, all of the infrastructure needed for that critical IT load. We went to the market with a trial balloon of $10 million a megawatt to buy or $1 million a megawatt a year to lease. Both of these numbers represent a discount to where publicly traded comps are. Whether that's the correct price or not, ultimately, the market will tell us. We invested about $100 million in almost a year in doing these conversions. The number of facilities that we converted is far larger than what we had initially targeted.

So on day 1, when we announced the transaction in September of '22, we thought we would convert 45 of the 482 buildings. We actually have grown that to 125. And then the amount of power that was going to be converted went from the initial 45 to about 150 megawatts, of which the 109 has been viewed as surplus to Cogent's core business. The second asset that Mike touched on are our inventory of IPv4 addresses. These are the unique hexadecimal numbers that make the majority of the Internet work. The world ran out of those addresses in 2017. We have been leasing addresses. We have raised prices on leases and we have continued to lease out increasing volumes. We took a portion of that lease revenue and securitized it with an asset-backed securitization that has raised $380 million and we still have about 23 million completely fallow addresses. We are considering selling them, perhaps leasing them, perhaps doing more securitization.

And then finally, the Sprint physical network is comprised of anywhere from 24 to a total of [ 144 ] strands on any given point-to-point path. We do not need all of that inventory to run our IP business, to run our wavelength business and to account for future growth. So we've also considered and we've done just a handful of small dark fiber sales but we will consider those going forward. Each of these noncore assets represent things that can be sold off to help strengthen our balance sheet. They are not a recurring revenue. The key value creator at Cogent is the recurring revenue and operating margins of that ongoing business. Cogent will be able to grow its total top line revenue based on those 3 categories I described earlier at between 6% and 8%. We will be able to deliver 200 basis points a year of margin expansion. And we'll do that with declining capital intensity. Those factors will allow Cogent to generate increasing amounts of free cash flow per share that can be returned to shareholders.

Great. I just wanted to see if there are any questions from the audience before I move on. Dave, maybe you can talk a little bit about that declining capital intensity point. You're expecting to reduce capital intensity, I think, to about a $100 million annual run rate later this year. How do you balance that piece with the need to invest in the business to compete?

So it's actually a little more than [indiscernible] if you looked at principal payments on capital leases, which is another form of CapEx. It's just a different accounting treatment. And together, those 2 numbers are about $140 million. So the capital intensity of Cogent prior to acquiring Sprint was the lowest in the industry with the highest growth rate based on the network architecture that we deployed and the product discipline that we applied to the network. When we acquired Sprint, our capital intensity increased for 3 reasons. We had to repurpose the Sprint network. We had to refurbish the data centers and we had to interconnect the networks. That work is all behind us. And with that, we will be able to have lower capital intensity in our IP business and lower capital intensity in our wavelength business. So one of the key differences in our architecture versus our competitors is it was built solely to deliver wavelengths, not other services.

In the deployment of transponders, nearly 40% of all transponders manufactured never become productively deployed. There's stranded capacity. The architecture that we have deployed is far more effective at reducing that amount of stranded infrastructure. It is a very different topology than the companies that we compete with. That's what allows us to install faster and what allows us to get better transponder efficiency.

Maybe just in closing, maybe you could just talk a little bit about capital allocation. The company has consistently reiterated its intent to sequentially grow the dividend. You're obviously in the midst of several potential asset sales. Talk a little bit about capital allocation, your willingness to do additional M&A to the extent that the conditions make sense.

I think on M&A, it is highly unlikely. Between 2005 and 2022, we looked at 825 markets and did a grand total of 1. So I think we've been disciplined. I think that will continue to be our mantra. That doesn't mean there won't be things that come available. But right now, there's nothing that we're actively looking at. Two, our leverage has peaked. We knew it was going to peak in Q2 of 2025 due to the investments in the Sprint network and the reduction in subsidy payments from T-Mobile. We are going to consistently delever from this point forward. On a net leverage basis we're about 6.6x levered. That is above where we target and we do need to delever. We have bought back 10.9 million shares or nearly 23% of our cumulative outstanding float.

We did some buybacks in the second quarter. And we also have a dividend that we've grown sequentially for 52 consecutive quarters, quarter-over-quarter. We believe we can do all of the above. There is a debate on what is the appropriate mix of dividend, buyback and delevering, particularly as the decoupling of our share price from our dividend yield, where historically, we had a much more market normal dividend yield. Today, we have one that is clearly telling us the market doesn't believe our dividend.

Well, we're out time. So we'll wrap it up there. But Dave, it's been such a privilege to have you on stage here with us. Thank you for joining us and thank you, everybody, for listening in.

Thank you all and thank you, Mike. You did a great job for literally just initiating. Thanks.

Thank you, sir.

So once again, everyone, thank you for coming out and special thanks to Dave Schaeffer. Dave, great to see you again. It's good to be covering the space and get reconnected with you and Cogent. So thank you once again for attending our conference.

Well, thank you. Hey, Ana. Welcome back to telecom from software. Hopefully, it's a little more exciting. And as always, I want to thank Bank of America for a great venue, the opportunity to present. And I'd like to thank all the investors for taking time to hear a little bit about what we're doing.

Yes. Absolutely. Look, given a chance to catch up on what's Cogent, what's happening at Cogent and yourself has certainly made it more interesting for me as I've ramped back up on the space. So I want to start by going back to last quarter and then maybe connecting some things to some previous guidance you gave about the full year exit rates.

So if we can start on the wavelength business, and I'm going to paraphrase my interpretation, and then please correct me if I'm wrong. But the installations in 2Q came in below Street expectations. You said that, yes, but the backlog, right, is very large and installations were lower because customers accepting delivery was less than you expected, and that was due in part because they're so conditioned to not expect -- or to not get delivery in the time frame that you were delivering, right?

But I think your commentary remained positive or you reiterated your exit rate expectation for the business for 2025. And I hope that I caught all that right. If I didn't, please correct me. So can you take that as a question and just kind of fill in what I've missed and connect the 2Q with the exit rate for 2025?

Yes. So the wavelength business is a new business for Cogent and also a new business with the Sprint network and Sprint assets. So when we announced the acquisition of Sprint 3 years ago actually at this conference when it was still in L.A., our plan was to wave-enable the former Sprint long-distance network, and we laid out a time line and a path to do that.

Probably the biggest disappointment during that process was our inability to sell a significant number of wavelengths between large data centers. So when we initially laid out a target of connecting 800 data centers, we said that would take us almost 2 years to complete post closing. In fact, we did it in 17 months.

And during that process, we concentrated our initial work on the largest data centers. While we did sell a small number of wavelengths and installed them, the vast majority of the wavelength demand tended to be from smaller data centers to larger data centers. And at the end of 2024, we had fully enabled 802 sites to be able to deliver waves at 1 of 3 speeds, 10 gig, 100 gig and 400 gig and do that delivery in 30 days.

This is an industry where waves have traditionally been installed on an as-needed bespoke basis with generally 4 to 6 weeks to get a firm quote and then another 2 to 3 months for install with a significant number of waves that are quoted never installing due to issues with the service provider and their network availability.

In Q1, we installed a significant number of waves, but virtually all of those waves were installed at the very last day of the quarter. And for that reason, there was a significant disconnect between our revenue and our unit number of wave installs. We have not yet implemented a forced billing model, even though we have the right to do that for customers. Because we are a new entrant in the market with less than 1% market share, we have treaded lightly on pressuring customers.

In the second quarter, we installed significantly more waves than we ended up recognizing revenue for. And I believe that, that gap will close over the next several quarters as we validate for customers our ability to actually meet the delivery windows that we have laid out. Even with that, albeit from a very small base, our revenue in the wave business grew sequentially 27% and 149% year-over-year.

Now our run rate in Q2 was a little above $36 million annually. And while the company does not give annual guidance in validating the reasons for the acquisition, we laid out a series of multiyear targets that included growing the wavelength business to $500 million in a period that would be 5 years post closing, and that would result in mid-2028 doing $500 million in wave revenue.

We have built a significant funnel of wave opportunities. When we announced the transaction, we were expecting to sell to 3 customer bases: international carriers, regional carriers and content distributors. A fourth group of buyers emerged, which quite honestly was not anticipated in September of 2022, which is AI training. And we think that will drive incremental growth for the aggregate market and help accelerate our ability to meet our revenue targets.

We have issued a number of KPIs that are designed to give investors some points to measure Cogent's progress. But ultimately, the only KPI that should matter is our GAAP reported revenue. And I think over the next year or 2, we will migrate away from these qualitative KPIs and focus on those GAAP numbers. We expect the cadence of installs to pick up, and we expect the lag between install and customer acceptance to shrink.

We believe that the experience we've had in selling high-capacity Internet and data centers is indicative of what we expect in the wavelength market. So while it is true the unit numbers were below expectations, the revenues were not materially below what we expected.

Okay. And just to boil it down how I'm thinking about it, Dave, is you laid out that customers were conditioned to you said kind of 4 to 6 weeks for delivery and then sometimes it was never delivered. And so one thought that I had asked you last quarter was, do you think maybe that these customers were just over-purchasing, right? Just that would be a rational thing to do if their experience was that it takes a long time and sometimes it's never delivered.

And then have you seen an improvement in installations, right? Have you seen an improvement in customer acceptance this quarter versus 2Q? Because I think to hit your target for your exit rate, you would have had to do something like 1,000 installs a quarter in 3Q and 4Q. So have you seen improvement? And is my hypothesis about customers overbuying, is that on or off base?

So two very different questions. We have seen zero cancellations before install. So that indicates that customers were not overbuying, but rather were caught off guard, in part because they were conditioned by our competitors to believe that we could not install and that the quality of our service would not be what it has turned out to be.

I think the fact that we have installed in 428 unique locations at the end of the quarter and have installed services now for several hundred unique customers is helping us get the opportunity to bid on a larger portion of those customers' backlog of orders. There are approximately 9,000 waves per month that come out of contract with the installed suppliers because most of the waves are going through a capacity migration from 10 to 100 gig and a small segment from 100 to 400. That means there is a new buy decision required in each of those waves as they come out of contract.

So I feel that the breadth and depth of our funnel is a good proof point that we will hit metrics that we have laid out both in the near term and most importantly, in the longer term. So I think that we will continue to see the differential between install and customer acceptance shrink. But the sample set is too small, and we are too early in the wave business. What is maybe a little hard for investors to fully understand is while this business is important and it is the main justification of acquiring the assets from Sprint from T-Mobile, it is a brand-new start-up in the public arena.

So within Cogent, waves represent less than 2% of our revenue. It is a brand-new product set and one that we have to demonstrate credibility with customers. I think the fact that our ARPUs went up and our discounting has not been as extreme as we expected it to be at this point is also a good indication of our ability to capture market share while maintaining price discipline.

Okay. I'd like to double-check sometimes that I'm hearing or understanding things correctly. So what I hear is that the longer-term opportunity is as large or larger than you thought when you initially did the Sprint transaction. There are a lot of customers coming off contract that create new growth opportunity for Cogent, right? That hasn't changed.

Maybe some of the near-term metrics or KPIs or commentary that you gave to be helpful when you closed the deal are less relevant and might slip a little bit because it's a newer business, and it's hard to project growth and slope of growth early on in new business. But the longer-term outlook is intact and still very positive. Is that -- am I encapsulating all that correct?

I think you are, Mike.