Bank of America. I cover telecom and technology and the fixed income research side. This is our Leveraged Finance Conference. Thanks to everyone joining us, and we are thrilled to have Cogent Communications and Dave Schaeffer, the company's Chief Executive Officer and founder and chairman. Dave, thanks very much for being with us.
Ana, thank you for hosting me. Thank Bank of America for a great venue, and as always, I want to thank investors for taking some time out to hear a little bit about Cogent.
Okay, so there's always a lot going on with you in Cogent. So I'm ready to start diving into questions, but any opening remarks you'd like to make or?
No, I think let's just hit Q&A.
Okay, so I thought we'd start with just a status update on the initiatives around the Sprint Wireline business, which you acquired. And just to remind everyone, that was May 1, 2023, so a little bit over about a year and a half now since you made that acquisition. So just start with the synergy status. So the original target was $220 million of annual cost saves in three years. So that would be by May 2026. But as of the quarter you just reported, so 3Q, you said you achieved $165 million, or actually 75% of those cost saves. And I think there's been headcount reductions. You've also worked to exit unprofitable contracts and leases. So of the remaining $55 million of the original target, what's left and how will those roll in from a timing perspective?
Yeah, so two points. First of all, it is likely that we will take that target up at some point as we continue to identify additional areas of cost saving.
Are you ready to sort of size?
We are not ready to put a number out today, but we are comfortable in saying it will be greater than the $220 million that we outlined. Some of those savings are governed by contracts that we acquired, and we have to honor those contracts. Those could either be to suppliers or to providing service to customers for services that are unprofitable and in many cases gross margin negative. We will continue to focus on G&A savings, such as headcount reduction, but most of those savings have now been achieved. The remaining savings will come from certain lease and other obligations rolling off as expected, and then secondly, the ability to continue to migrate customers off of off-net facilities to on-net. Just to remind investors, when we announced the transaction, it was actually at a Bank of America equity conference in September of 2022.
The acquired business had $565 million of revenue and negative $300 million of EBITDA. In the intervening nine months from that announcement till we closed, T-Mobile initiated a number of cost reduction exercises with our consent and in working with them. We were able to shrink the business to $490 million in revenue and negative $190 million of EBITDA. We have targeted to be EBITDA break-even in that acquired business by May of 2025. We are on track to beat that target as well. Our main areas.
May of what year?
May of 2025 to be break-even. So it was two years post-closing and in three years post-closing to be generating approximately 20% positive EBITDA margin on the acquired business. And again, just to refresh investors' memory, this transaction actually had two separate components. It had the acquisition of the operating business selling to mostly large enterprises a variety of products, 28 products in all, most of which were not strategic or subscale and actually gross margin negative. The second part of the transaction was our ability to acquire the physical network that Sprint had built and was sitting dormant. That network was comprised of 19,000 route miles of inter city fiber, 1,200 route miles of metropolitan fiber, and 482 telco buildings representing 1.9 million sq ft and 230 megawatts of power. That infrastructure was built at a capital cost of $20.5 billion between 1982 and 1991. We were able to acquire that for $1 as part of the transaction.
Okay. And a big part of that, your rationale was your ability to enter the wavelength market, right? So I think you've been pretty clear that while you've been seeing a lot of demand for selling wavelengths, provisioning has been more complicated potentially than you originally anticipated. So your wavelength revenue has been growing, I think, about $1 million a quarter. So it breached $5 million in 3Q24. But what you've talked about as instituting a streamlined internal process, a provisioning process, which I believe is supposed to go live around January 1 or be complete by January 1 of 2025. And I think you've said that you believe you'll be able to provision a wavelength faster than anyone else in the industry. So, I just wanted some background on that and an update on the status if you're kind of ready to go live as of the beginning of the year with that.
Yes. So first of all, we will be able to provision wavelengths in 800 carrier-neutral data centers across North America in January of 2025. We also will be able to shrink the provisioning window to ultimately a two-week window, probably in January, a little bit longer, maybe three or four weeks as we ramp up our provisioning systems. And we have sufficient resources in hand to be able to provision about 500 wavelengths per month. We have about 3,400 wavelengths in our backlog as of the end of Q3, and we had provisioned approximately 1,100 wavelengths representing an annualized run rate of about $20 million from zero, but clearly a long way from the $500 million run rate and wavelength revenue that we anticipate by May of 2028, five years from closing. In order to accomplish this, we had a number of foundational steps that we had to undertake.
We had to physically extend the Sprint network to touch our metropolitan network and 110 markets. That work was complete in February of this year, February of 2024. We also needed to deploy transponder shelves in those 800 data centers to accept wavelengths. We needed to reconfigure our metropolitan networks to optimize them for wavelengths and to segregate the IP services from the wavelength network. And then we needed to deploy reconfigurable add/drop multiplexers at the intersection of the two networks, metropolitan and long haul. In architecting our network, we took from our 20 years of experience in IP. So in many ways, Cogent is synonymous with low pricing, and that is a key part of our strategy and has allowed us to become the largest carrier of internet traffic in the world. We carry approximately 25% of global internet traffic.
And we do that because we underprice our competitors at more locations, but also we architected an IP network that cut the provisioning time of a high-capacity transit circuit from 90 days to nine days. And we consistently win third-party awards as the fastest and most reliable company to provision services. We're able to do that not because we're particularly special, but because we built a network that was optimized for IP, taking that learning and then projecting it into the Sprint network. The Sprint network was not designed to carry IP or wavelength services. It was a voice telephone network. It had sat dormant since 2015 with only 7% of Sprint's revenues touching its network. And we took that empty network and said, can we take a portion of the capacity, a pair of fibers, and build a wavelength-optimized network?
What that entailed is getting a network where to turn up a wavelength, it would only require two field visits and no pre-engineering work and no post-installation engineering work. We have architected that and substantially completed that, and by the end of this month, that work will be completed, allowing us a much more rapid provisioning time. Today, the industry is characterized by companies that take between 90 and 120 days to install a wavelength, and those wavelengths are built on a custom basis where each one is engineered and then provisioned on a one-off special basis. The number of permutations of wavelengths is truly astronomical. With 800 data centers where you can have an any-to-any wavelength provision, there are 4 times 10 to the 1,978th power number of combinations.
It is impossible to pre-position that kind of capacity, but you can architect the network in a way that the only steps that are necessary are the deployment of two pluggable transponders at the endpoint, and everything else can auto-configure. That is impossible if you are co-mingling other services on that network. So what Cogent will be running is its traditional IP network for our core services, which today account for about 91% of our revenues, and then we will have a second network that is optimized for wavelengths that will allow for this rapid provisioning with three speeds, either a 10-gigabit wavelength, a 100-gigabit wavelength, or a 400-gig wavelength. We have built the sales processes and tools into our CRM system, so when a customer inquires about a wavelength between two North American data centers, there's a drop-down menu of centers. The salesperson clicks on the two centers. It auto-generates the physical path through the network, a map with latency accurate within one hundredth of one millisecond, a KMZ map or a detailed geographic map of the network with accuracy of one meter, and the pricing for that customer. No other vendor in the industry can provide this level of detail even with that 120-day installation.
Okay. So super helpful with regard to the level of detail there. But then that takeaway is that you believe that you both have the demand and the provisioning capacity now where that so your goal to reach $500 million of annual wavelength revenue in 2028 was effectively like three and a half years from now. You believe you're on track to go from $20 million currently to $500 million. And we should start really seeing that ramp in the first quarter of 2026.
That is correct, Ana. And we have confidence in looking at the total addressable market two ways. It is the global wavelength market's about $7 billion. The North American component of it's $3.5 billion. Of that $3.5 billion, $1.5 billion is metropolitan, $2 billion is intercity. We think we can achieve similar market share gains in wavelengths as we have in IP, i.e., 25% share. That's a top-down view. On a bottoms-up basis, we looked at 287 wholesale sales reps that we have, probably the largest sales force focused on that market in the country. And we queried them on the accounts in their deck and the number of wavelengths each of those accounts could generate and got very comfortable with our $500 million target by mid-year 2028.
Okay. Could you talk about just what the margin contribution, EBITDA contribution, you know how we can expect from this ramp is?
Very high contribution. So these are on-net services with minimal amount of capital, meaning the only CapEx that we will be deploying are those optics at the two endpoints, and the contribution margins will be above 90%.
Okay. So I know you don't typically give annual guidance, but if you want to give it, we'd be happy to hear it. But so as part of the deal that you had with T-Mobile, they are compensating you through an IP transit contract, which they're not utilizing as far as I understand. And because they're not utilizing it, you can't book it as revenue, but you do take the payments and include them in your EBITDA because I don't want to call it a subsidy, but it is a compensation for the loss. So we put together so right now, and then I think for the next 13 quarters, you're going to be getting about $25 million a quarter. It's $100 million a year.
And you need to grow out of the negative EBITDA from Sprint, which I think you said you're on track to do by mid-year effectively, right? And then you should be growing into incremental EBITDA from the wavelengths. Is that the right way to think about how you're going to be growing out of this compensation, which is currently benefiting your EBITDA to the extent of $100 million a year?
It is temporary. A couple of clarifications on the points you've raised. First of all, as an inducement to take over this money-losing business from T-Mobile, in addition to selling us the network assets for $1, they gave us $700 million in the form of a transit contract. They are using several percent of that capacity, but it is not a sufficient amount of capacity for us to record it as revenue. Because these are cash payments coming in, we are including them in our EBITDA calculation. Prior to the acquisition for full year 2022, Cogent did $600 million in revenue with about $260 million of EBITDA. The classic Cogent business of selling IP-based services has continued to grow and has had margin expansion.
But layered on top of that was the initial negative 190 that we got at closing and now down to well below $80 million, heading to zero by May of 2025. To offset that, we received $350 million of the $700 million payment stream in the first 12 months. So $89 or $88 million a quarter for the first year, stepping down to $25 million a quarter for the next 42 months. And as a result, our EBITDA stepped up with the acquisition from $260 to $352 million in 2023 and will be similar, slightly below that, but similar in 2024. Because the full-year subsidy payments are declining by about $100 million year over year, we need to keep EBITDA even, make up that difference through a combination of synergy savings and growth of high-margin products. So within the acquired operating business of Sprint, we have done three things.
We have eliminated non-core products to the extent we could that were gross margin negative. That has helped us improve margins. We've eliminated G&A expenses, and then finally, we've moved as much traffic from off-net to on-net as possible. And that process is continuing. We expect our EBITDA for 2023, 2024, and 2025 to be effectively flat, even though on a year-over-year basis, 2023 to 2024, we had a $100 million headwind that we've made up for, and then 2024 to 2025, we have another $100 million headwind in these lower payments that we will make up through these savings. Then as we think about 2026, we start to see material growth in EBITDA because the payment stream from T-Mobile remains constant for 2025, 2026, and 2027. But the underlying growth allows our EBITDA to grow. You are correct.
We don't give specific annual guidance, but we have guided investors to expect that five years post-transaction, so that is May of 2028, the company will be doing $1.5 billion in revenue up from the $1.1 billion today. And EBITDA will be up from $350 million to $500 million. And by that point, the subsidy payments from T-Mobile have completely gone away.
Your word. I didn't say subsidy. Or maybe. So your word, not mine. I call it compensation, but got it. Okay, so we are, as usual, time's flying. I do want to hit on capital allocation and capital structure strategy because we are at a credit conference, so let me just ask the questions and you could decide the best order to answer them, so first of all, just on your dividends. So you have increased your dividend for 49 consecutive quarters. I have that correct. It's pretty generous dividend. So it's almost a dollar per share per quarter, so $99.55 annualized, about almost $200 million, but your free cash flow right now is constrained in part because you're actually, we haven't had a chance to talk about it yet, but you do have a data center kind of investment project.
You know one of the things we've learned in telecom is it can be capital intense. You know sometimes the best use of capital is to reinvest in the business rather than paying it to shareholders. A couple of questions. Do you expect to generate enough free cash flow to cover the dividend? And have you thought about at least stopping the increases in the dividend?
So we consider the dividend and its increase every quarter. We have no plans to either stop the dividend or stop the increase in the dividend. So I'm going to go back in time. Cogent began returning capital to shareholders in 2006, 2006, excuse me. We were very lucky. We were able to build the company with no net debt. In 2010, we added leverage with your help and Bank of America's for the first time and have slowly ratcheted our aggregate leverage up to about a net of $1.5 billion. For 15 years, we have returned more than 100% of free cash flow. So this is nothing new. We are using the balance sheet as a strategic asset. Our leverage had hovered between two and a half and three and a half times up until the pandemic.
We were raising our dividend $0.025 a share a quarter sequentially. When the pandemic hit, our growth rate in our corporate business moderated significantly. We waited too long, and our leverage actually breached 4, went up to 4.2 times. With the acquisition of the Sprint business from T-Mobile, we rapidly delivered to a trough of 3.17 times levered, so well back within the stated range we had and delivered. As those subsidy payments dropped off, our worst quarter on an LQA basis is last quarter because of the step down in payments, and on an LTM basis, it'll be mid-year next year. Our leverage will go up and probably breach 5 times. We still have confidence that we can grow the dividend and continue to invest adequately in the business.
Different than many telecom companies, we for the past roughly 18 years have decided to return capital to our equity holders using debt as a tool to help us do that. That strategy has not changed. The growth in the business is dependent on our ability to monetize the assets we have, not to invest more capital and perpetually grow without generating returns on capital above our cost of capital. Now, what we have done is looked at our asset base, and initially, we had not planned to spend any significant capital on our data center footprint. We had a pre-existing data center footprint of 634,000 sq ft and 69 megawatts. We acquired these 482 buildings. We identified 48 of them as suitable for conversion at data centers. We still did not need extra capital to do that. That work has actually already been complete.
We today have 104 total data centers operational. But we have approximately 1 million sq ft and 100 megawatts of power that we had expected to sit fallow. With the acute shortage of power driven by AI, we elected mid-year this year to start to refurbish that excess capacity. It will cost us nearly $100 million to do that over a 12-month period, but it should result in 100 megawatts of surplus available power that we can either lease or sell on a wholesale basis. We are in active discussions with counterparties and expect to be able to monetize that footprint sometime in 2025.
You've talked about raising debt to help fund the dividend. This last year, you issued ABS on your IPv4 portfolio, $206 million in April, and I think you've since said that you've increased revenue so that with these price increases by 50% run rate, so your borrowing capacity in that ABS vehicle is now increased by 50%. Is that?
More than 50%.
How much is it? Is it $100 million of borrowing capacity? Can we expect you to tap that?
Yeah, so it is a master trust, so it allows us to borrow incremental against it. We chose only to take the A1 tranche, which was double B rated. We raised $206 million against a revenue tape of $33 million. That revenue tape today is up to about $51 million. That is through both a partial increase in prices for the base. We have not increased the entire base yet, and through incremental sales of addresses in the trust, leasing of those addresses. We will have the ability to borrow probably up to another couple of hundred million out of the trust. I don't think we will use all of that capacity, but probably will do an incremental draw on the trust sometime in 2025 to give us some extra liquidity at the lowest cost of capital. Also to remind investors, we have two tranches of high yield.
Our secured high yield comes due in June of 2026. It is a 3.5% coupon, so we're in no real hurry to refinance that early. Then we have $700 million of unsecured debt at a 7% coupon that comes due in mid-2027. So I think as investors think of the cadence of our debt raises, probably a small incremental draw against the ABS mid-year, followed by a refi of the secured debt, probably by end of 2025 or early 2026, potentially with an expansion of that based on our EBITDA staying within the covenant guidelines and RP limitations that we have. And then finally in 2027, doing a similar refi of the $700 million unsecured.
Okay. With that, we're out of time, but that is a perfect roadmap, I think, that you gave to the credit investors here. So there's a lot of stuff we didn't touch on. And I think it's maybe the first time that we actually did not touch on the core business or the legacy. There's a lot going on there, but there's certainly so much going on. So I think you gave a great overview, and I appreciate you being.