My name is Phil Cusick. I follow communications and media space here. We're gonna welcome Dave Schaeffer, Founder and CEO of Cogent Communications. Dave, thanks for joining us.
Hey, Phil. Thank you for hosting me. Thank J.P. for a great venue, Most importantly, thank all the investors for taking some time to hear a little bit about our business.
Excellent. Well let's start with the recently closed acquisition of Sprint's global media business. This is the first M&A deal you've done in nearly 16 years. T-Mobile's been trying to sell it for what seems like almost that long. Why this one and why now?
Well, it has been a long time, we've looked at over 825 potential targets and chose this asset. I think there are three reasons. First and foremost, we think there's an asset that can be repurposed and create significant value. Two, there is a customer base that has not been tended to, many of the products and services that they are purchasing are no longer economically viable. There's an ability to improve that business by grooming the product set, modernizing the services we deliver, and continue to support those large enterprise customers that have a 30-plus year relationship with Sprint. I think the third reason, to be blunt, is we were paid $700 million to take it.
Rarely is there a situation where an asset 20 years previous was in the market and under contract to be sold for $129 billion, and then 20 years later, the seller is paying the acquirer $700 million. I think they're de-risking the acquisition for us.
How does the size of this company in terms of whether it's fiber or route miles compare to the existing Cogent?
Yeah. The current Cogent network is approximately 62,500 miles of inner city fiber. The Sprint network is 19,000 route miles. Cogent's metro network is 18,200 miles. Sprint's is 1,100 miles. It is smaller in terms of geographic reach. In terms of traffic carried, Cogent's network is carrying about 1.1 exabytes a day. The Sprint network was carrying slightly under 10 petabytes, so roughly about 105 times bigger the Cogent network. By revenue, Cogent was approximately $600 million in revenue. The Sprint network at time of deal signing was $560 million, at closing $470 million. A significant acquisition in terms of revenue scale and network.
The real opportunity lies in the ability to take what is effectively a fallow footprint and begin to generate revenue off of it. This was the first nationwide fiber optic network. It was built in the mid-eighties. It terminates at a number of tandem switch sites that were purpose-built that are going to be repositioned as data centers. While we are being paid to fix the operating business, the real reason for doing the transaction was to control the assets and begin to sell new products.
It seems like a very underutilized asset, but it's not like Sprint didn't need money for the last 10 years. Why weren't they going after the Wave market as well?
We'll turn the clock back. This is actually an asset that Cogent had attempted to acquire twice previously. In 2013, when SoftBank was looking to acquire Sprint, we were engaged by their then current CFO, Joe Euteneuer, met with SoftBank, and actually agreed on the economics of a deal that were much more favorable to the seller. The business was in better shape. Eventually that deal was not consummated. Fast-forward to 2018 when T-Mobile acquired the asset, they engaged a banker, ran a process. The business had deteriorated. We had, again, bid on the asset and were rejected, as were other bidders. The business had deteriorated further. At that point, I think T-Mobile was not willing to accept either no cash compensation or pay a purchaser.
After evaluating the business, they hired a consultant, they looked at the asset and had attempted to implement a Wave strategy upon a recommendation of that consultant. They spent about $10 million. They generated about $8 million in annual run rate and concluded that it was going to be too hard for three reasons. One, their network terminated in very unusual locations, so they would have to spend multiple billions of dollars to extend that network into the major carrier neutrals and get the reach necessary to satisfy the wavelength market. Secondly, they had no sales force that was equipped to sell wavelengths, so they would have to develop and hire that. Finally, internally, they did not have a management team that was familiar with this market and wanting to execute the business.
Putting those three concerns together and looking at the potential scale versus their wireless business, they concluded it was actually better to either shut the business down or attempt to sell it again. Prior to our acquisition, T-Mobile had hired yet a different consultant and went through an evaluation and concluded that a shutdown was most likely at a cost to the company of about $1.5 billion. We stepped in when they engaged a different banker. The transaction with Cogent includes a combination of or $700 million of direct cash payments to us, $25 million of severance payments made to us for employees we terminate, $100 million of additional indemnification. Finally, about $500 million in expenditures by T-Mobile between signing and closing.
The total cost to T-Mobile was slightly less than their $1.5 billion shutdown cost at about $1.35 billion. I think what they got with Cogent was three advantages they wouldn't have gotten with a shutdown. One, they got certainty. Two, there were no regulatory entanglements. Finally, there was no brand damage as we agreed to support all customers going forward.
Excuse me. The $8 million of Wave revenue that came with the business, that was all sold by T-Mobile a few years ago. You've ramped up that pipeline pretty quickly after the announcement of the deal. How should we think about that, and what's been your ability to close those since the close of the deal?
Once we announced the transaction, we entered into a commercial relationship prior to getting regulatory approval that allowed us to go to market and sell wavelengths as a reseller of T-Mobile. As a practical matter, none of those wavelengths were really installed, and we did not expect that, but it allowed us to test market demand. We have generated several hundred Wave orders. The transaction just closed May first. We are in the process of provisioning those initial orders. In the Wave market, the traditional window for provisioning ranges between 90 to 180 days, in some cases more. Cogent has busily worked at connecting the Sprint network to carrier-neutral data centers and has the ability to sell wavelengths today in 200 of the 800 facilities that we have on our network.
Over the next 18 months, we anticipate being able to sell wavelengths in all 800 U.S. carrier-neutrals on our metro footprint. We also expect to be able to collapse that provisioning window from the market average of 90-180 days down to our traditional on-net window of a guarantee of 17 days and an actual observed install of nine. We will begin to rapidly provision those services in those first 200 locations, probably with an installation window of between 60 and 90 days initially. We are busily working on completing the interconnection of the network and standardizing the network architecture to allow for more rapid provisioning.
you know, what some of the incumbents in the Wave market might say is, "This is old fiber. It's not necessarily what customers want today." What's the response been from customers as you go out and sell this?
It's actually been very positive. There are some unique attributes to the network. It is absolutely correct that the network was the first nationwide fiber optic network built in the mid-1980s, so it is a 40-year-old asset. It was built with SMF-28 fiber. That fiber was initially viewed to be end of life in the mid-1990s, and the market shifted to non-zero dispersion shifted fiber.
I don't know what that means.
It's two different ways in which the light moves through the fiber. A light pulse, when it goes through the fiber, tends to pull apart, chromatically separate, becomes blurrier. In a non-dispersion shifted fiber, the fiber has chemicals in it that kind of push the light pulse back together. That was necessary when the transmission medium was non-coherent. All of the initial fiber optic deployments were done using non-coherent, meaning pulses were sent individually down the fiber. In the mid-2000s, as the equipment vendors struggled to get greater throughput per wavelength, all of the vendors, Ciena, Cisco, Nokia, Fujitsu, NEC, all of the vendors migrated to coherent technology, meaning the information is sent in clumps down the fiber. The non-zero dispersion-shifted fiber that was put in the mid-nineties through mid-2000s actually is worse because that compression of the signal actually distorts the coherent transmission.
The older fiber is actually better. The second reason there's an advantage is this network was built by a single provider, meaning there are very accurate records of where the network exists. Our competitors are roll-ups that have had fiber built by dozens of independent companies, and therefore do not have accurate records of their outside plant. We can auto-generate a file with every proposal that end-to-end is accurate instantaneously with 1 meter of accuracy. That's very valuable to a buyer, particularly since the wavelength market is about deterministic routes, whereas the Internet is indeterministic. It's path independent. The third advantage is that the fiber was deployed differently. It was in an armored cable, typically deployed on railroad right of way, buried 6 feet deep. Most newer fiber is in plastic conduit only 2 feet deep.
Even though the fiber is 20 years older, it's had less cuts per kilometer, typically by almost an order of magnitude than the fiber that was deployed 20 years later. There are less loss in the fiber, less splices. The fiber in 90% of the routes is unique. No one else shares that route. In building a deterministic network with wavelengths, having unique routing is extremely important. We have the best of 3 possible combinations. High quality fiber for the highest transmission, unique routing, and a maximum number of endpoints. Couple that with the fact that we have a negative cost basis in the asset, we are in a superior position to our competitors in selling in this market.
Customers accept this argument?
Not only do they accept it, we quite honestly were a bit shocked by their desire to buy. If you go back and look at the wavelength market, it initially was dominated by AT&T and Verizon. The primary customers were LD resellers who put switchers in markets and bought wavelengths to connect those switchers. That market existed from the mid-80s to about 2000. In the early 2000s, the market changed, and the entire customer base became regional ISPs for wavelengths. AT&T and Verizon exited that market. Lumen and Zayo came to dominate that market. That independent ISP market disappeared. Today, the buyers of wavelengths fall into 3 major categories. Content generators who are linking together major content aggregation sites. A Facebook, a Google, a Netflix, a Microsoft, even a content delivery network like an Akamai.
The second base of customers are regional access networks that are linking together islands of service. Someone like a Mediacom or a Cox or a Spectrum that have disparate markets that are linked together in a holistic market. Finally, there is still a corporate wavelength market. It's the smallest of the segments, and it's dominated by very large companies like J.P. Morgan. All three of those segments have expressed interest in having a new supplier, unique routes, and are frustrated with the provisioning time and lack of endpoints that the competitors have. I think we are filling a gap in the market. Remember, Cogent already has existing transit relationships with three-quarters of the potential buyers of these services. We also have the largest sales force in the industry focusing on this segment.
You know, we are, depending on how you measure it, the number 1 or number 2 transit provider in the world. It's off of these transit relationships that we're now going back to the same customer base and selling them a adjacent product with the credibility we've established in gaining market share in transit.
We talked about not selling that at a discount. Could you sell it at a premium to the current market?
In transit, we became defined by undercutting the market and becoming the price leader. We had to do that because there was little differentiation, and we had no credibility. I think here we will use price. We will clear with price, but I think it's not necessary to go in with the same aggressive 50% discounting to market that we did with transit. We will not sell at a premium. We will sell at market or a discount. Remember, this is an asset that we were paid to take. Even if we spend about 60% of the $700 million that we're getting from T-Mobile to fix the enterprise business, we have a negative basis in our Wavelength network that gives us the ability to price as aggressively as necessary in the market.
I wanna just finish up with the numbers and without going through all of the different numbers that you've given many times in the past. I think there's some confusion out there in terms of what the aggregate numbers are supposed to be on the combined company. I'll tell you, our estimates for 2024 are company revenue in the mid $1.2 billion and EBITDA in the high $200 million. Does that seem like a reasonable thing given what you've said publicly?
That seems reasonable, Phil. You might be a little low on the EBITDA number, but, you know, going in, Cogent organic was about $600 million, $250 million of EBITDA. We add on to that the $13 million a month payment from T-Mobile, so about $170 million will flow through to EBITDA. We start with a burn rate of negative $190 going to negative $80 within 12 months on the acquired enterprise business. We have roughly $30 million of commercial services to T-Mobile above and beyond the $700 million payment that are virtually all margin. You put those pieces together and come up with a combined run rate. The T-Mobile revenue, or the Sprint revenue from T-Mobile should stabilize at about $450 million run rate.
Mm-hmm.
The combined historic revenue and Cogent's revenues of $1,000,000,050, then you have to layer on top of the Wavelength revenue and the growth in Cogent's organic revenue. Your $1.2 million seems reasonable.
Okay. Okay, that's helpful. You know, let's switch to NetCentric a little bit and the deceleration on a constant currency basis in the Q1 down to 10%, was a little bit of a surprise. You talked about a lot of large customers bought traffic in the Q1 , that drove the price down. Why does that sort of shift on a quarter-to-quarter basis, and how do you think about it going forward?
Large customers tend to buy in big chunks. They are large. While we did continue to sell to a broad international base and smaller customers, the quarter was more dominated by a few large purchases. We have, I think, guided to the fact that our NetCentric revenues, these are the wholesale sales of transit services globally, 51 countries, 1,500 data centers, will grow at about a 9% growth rate. Going into the pandemic, we were actually substantially below that. We were growing at 3%. At the very beginning of the pandemic, that growth rate shot up to 26% year-over-year, fastest in the company's history. Its previous high watermark had been 24%, and that was when Netflix started selling, you know, streaming services versus mail order. We expect a reversion back to the mean.
It did decelerate to 10.2% year-over-year growth. We think that our NetCentric business will continue to outperform historical averages for the next several years for three reasons. One, more of the growth is international. Two, more of the growth on average is coming from a wider or smaller customer base. Third, traffic growth rates are still elevated due to the continued proliferation of streaming, particularly in international markets.
As you think about going forward that this is a little bit of a one-time thing, you think we can rebound a little bit? I think the year-over-year in the fourth quarter was roughly 16%, so we've been decelerating, but not quite as quickly as that 16% to 10%.
Yeah. Again, on the NetCentric side, we have guided to a long-term average steady state of about 9%. I do think since the pandemic, we've been substantially above that. The rate of deceleration was initially slower than we had anticipated. I think now we are asymptotically approaching a steady state, and we may actually see a few quarters of re-accelerated growth on a year-over-year basis. In general, the trend line should be to converge to about a 9% constant currency growth rate in NetCentric.
Okay. It doesn't sound like we should be looking for some kind of bounce back in the short term, though.
You know, we don't give specific quarterly guidance. Quite honestly, it's a usage-based service. It's difficult for even us to forecast beyond a day or two. It does appear that the gains in market share and outsized growth will continue in this quarter and probably the next several. There could be some balance. The long-term trend should be to model the historic average growth rate.
That's fair. Then in corporate, you know, we've talked about some initial excitement that maybe the corporate business would come back pretty quickly. That's taking a little bit longer. Any sign of improvement or deterioration as the economy seems to be sort of bouncing along?
I think you're too kind in saying it'll come back quickly. You know, our corporate business, where we sell to end users in the central business districts of major cities, we're selling them a fixed connection. That business had grown pretty consistently at about 11% for close to 15 years, with really only a couple of quarters of negative growth during the great financial crisis. During the pandemic, within 6 months, it went from being an 11% growing business to a negative 9% growth business. It has rebounded and reverted back to about a 0% growth. We're about halfway back from the trough of the pandemic, but we're definitely not back to where we were pre-pandemic. The path back has been longer and less linear than we expected.
We do believe that we will again revert back to a 10%, 11% growth rate in that corporate segment. The primary issue has been customers procrastinating and pushing off architectural changes. While the addressable market has shrunk due to vacancy, we went from a situation where the buildings were 94% occupied to 83% occupied. We only have about 20% of the building tenants as customers, so there's still plenty of addressable market in the buildings buying our services. Many of these companies have not yet decided what their final office and hybrid configuration will be, and until that becomes more certain, many of them are electing to just live with what they have.
As a result, the corporate rebound has been slower, and what we've tried to tell investors is it's probably going to take another year or two before it reverts back to being a consistent double-digit growth business.
I think what you've said is that you need customers to start making those decisions as to where their technology goes. We're a couple of years into waiting on those decisions. Are businesses starting to feel pressure on their old system, or can they run these for a few more years without a lot of trouble?
It's a combination. Some are, and that has been in part responsible for the return to basically flat growth. Others are willing to continue to procrastinate. There are really multiple catalysts. How many employees work remotely? Where do you want to have your employees? How many offices are you going to continue to support? How much of your traffic is going to be on a private VPN versus the public internet? How much of your compute and software will be cloud-based or SaaS-based versus perpetual license and on-prem? All of those factors are continuing to shape decisions, but we are seeing an improvement in the corporate environment. It's just not an instantaneous improvement. I think investors should expect to see steady progress.
There may be a quarter or two where it takes a pause. In general, you'll see a trend line of improving corporate growth.
Sounds good. It's a good place to leave it. Thank you, Dave.
Thank you very much, Phil.
Thanks, everybody.