My name is Brandon Nispel. I cover communication services for KeyBanc. This presentation is a 25-minute fireside chat. With us, we have Dave Schaeffer, the CEO of Cogent Communications. Dave, thanks for being here.
Brandon, thank you for hosting me. Thank KeyBanc for a great venue, and as always, I want to thank investors for taking time out of their day to hear a little bit about Cogent.
Absolutely. Let's, let's start on your 2Q earnings.
Well-
Give us the highlights, the lowlights, Dave.
Short conversation, since they're coming out on Thursday.
I kid, okay.
You could talk about them, I can't.
All right. Let's move on to the T-Mobile transaction. You guys closed on the T-Mobile transaction, I think, in middle of May or early May?
May. May first.
May first.
May first.
What are the sort of positive and negative surprises that you've learned thus far?
I'll start by saying, in general, the transaction is in line with what we expected. There are, however, some puts and takes. For the customer base, it was less pure enterprise, as during our due diligence, we could only see the top five customers by name, and the next 1,391 customers were masked. Once we unmasked those customers at closing, we saw that there were more corporate and net-centric type customers. Now, by revenue, it's still predominantly large enterprise, so I think what T-Mobile represented was correct, but by customer count, it was maybe a little more skewed towards some smaller customers. In terms of the network itself, it's the records, the quality of the infrastructure is as good, if not better, than we expected. I would say the general quality of the workforce is better than we had expected.
The quality of the information systems and some of the management tools is probably a little less than we expected. You know, it was a little shocking to us that there was not really a discipline around profitability, either by customer or by product. It's something that we're gonna have to work on. In general, when we look at the puts and takes, it's about as expected.
Got it. One of the things I get questions on, it came out in your 8-K, is, is the way that the T-Mobile contract is actually gonna be recognized. Can you maybe outline what you know thus far in terms of the payment from T-Mobile?
Yeah. What we know for a fact is we're getting $29 million a month in cash, generally around the second or third of the month following the month in which we issued the bill. T-Mobile is obligated to pay us $700 million for transit services. We have allocated ports to them. They are using a portion of that bandwidth, and the payments are equal for the first 12 months, totaling $350 million. The payments then step down to approximately $9 million a month for the next 42 months, and then in months 55 through 58, there will be a true up from T-Mobile for the working capital deficiency that we had at acquisition.
In general, we initially took the position, along with Ernst & Young, that the transaction should be counted as revenue under Section 606. T-Mobile and its auditors, PwC, thought it should be part of a bargain purchase transaction under Section 805. Because of the difference of opinion between the two companies and the two accounting firms, we went to the office of the chief accountant of the SEC and asked them for a ruling in the form of a pre-clearance letter. We did get that ruling from the Commission last Friday afternoon. We are busily pushing that through our documents. It is why we delayed the filing of our, our results a week beyond what we would normally file.
For co-investors who have followed Cogent regularly, we typically would have reported last Thursday, but instead, we're gonna report on the 10th, this coming Thursday.
Right. What's the difference to you, whether or not you recognize that contract as revenue? 'Cause cash is coming in regardless.
There really is no difference to us, and I think one of the things that was a bit surprising to the chief accountant's office when we had our oral presentation was, I represented to them, I truly didn't care what the answer was. I just wanted to be compliant with the way in which they wanted revenue to be, revenue or purchase gain to be recognized. If you think about the transaction, it really has two very different components to it. There's one part of the transaction where we pay $1 and acquire a series of physical assets, 19,000 route miles of inner-city fiber, 1,100 miles of metropolitan fiber, and roughly 1.6 million sq ft of owned technical space.... We paid $1 for those assets.
We, in fact, hired a third party, a Big Four accounting firm, to come in and audit that and give us an appraisal value, which we will represent as a gain. The second part of the transaction was acquiring the customer base of mostly enterprise revenue that was burning cash, and to mitigate that cash burn, T-Mobile entered into this transit purchase agreement over the 54-month period. They're really, in some ways, two separate forms of consideration for two different assets that were acquired.
Got it. Let's talk about the growth prospects for that business. I think, what we've heard is the most exciting aspect, maybe to you, is wavelengths, right? Where are you now? Where can you be in a year in terms of wavelength sales? We've sort of heard maybe the longer-term view of 25% share and $2 billion in revenue. What's the near term look like?
Yeah, the near term starts with a run rate at acquisition of about $8 million a year. That number should linearly grow to within seven years, being $500 million, or about $80 million run rate one year after acquisition, so the run rate on May 1st, 2024. From this acquired asset, there are really three things, again, that I need to discreetly answer your question. one, there is an enterprise customer base that will actually decline initially as we purge non-core products that are gross margin negative. Then we anticipate that acquired revenue stream to be flat, i.e., a zero-growth business, but one in which we can improve the margin characteristics to a positive 20% EBITDA margin in about three years. The second thing that we are acquiring are these technical facilities. We have earmarked 45 of the facilities for conversion to data centers.
When we convert these facilities, we will have about 160 MW of power and about 1.3 million square feet of sellable colocation space. We will sell that space both on a retail basis in one data hall and probably on a wholesale basis in some of the ancillary data halls in each of these facilities. In order to get these facilities to be sellable, we need to do three things. We need to connect them to the greater Internet. Two, we need to clear out literally tens of thousands of bays of dead equipment. Finally, we need to convert the power plants from negative 48 DC to AC 120. That process is underway in all of these facilities. The final thing that we need to do in these facilities is begin marketing that bandwidth.
In terms of the physical inner-city and intra-city network that we acquired, it is a network that is effectively empty. It has very little traffic on it. Our goal is to utilize that network to sell wavelengths. In order to do that, we need to connect that network to our carrier-neutral footprint in the U.S.. That's approximately 800 facilities. Two, we need to pre-stage equipment in the network so we can rapidly provision wavelengths. The industry average typically is a 90-180-day provisioning window. Sprint was in that range at roughly about 130 days. In the limited number of locations where we could provision, we've gotten that number down to 62 days. That is still unacceptable. Our goal is to guarantee a 17-day provisioning window and achieve an average of nine days.
That is exactly what we do in the sale of transit. When we look at how we will gain market share, it's through a uniqueness of routes, ubiquity of endpoints, speed and provisioning, and then lower prices. A little bit different than our entree into the transit market, this will be more driven by the qualitative attributes more than the quantitative lowering price.
Got it. I'm going to switch gears a little bit. Let's go back to your core business.
Sequential business-
I had more questions for you on T-Mobile, though. Corporate, let's start there. Corporate trends is, you know, -1% year-over-year. You've improved it from a trough of -8%. What's the outlook from you in terms of getting that business up to either a low or a mid-single-digit growth rate?
The first thing is, it's actually going to have very significant growth on a sequential and year-over-year basis due to- the acquisition of Sprint and something-.
You're talking, you're talking specifically for 2Q?
Q2, that's correct.
Got it.
That's both on a sequential and year-over-year basis. That's a one-time effect due to the fact that we acquired some corporate customers.
Understood.
We will report our numbers net of that reclassification, investors can see the underlying trend. Our growth rate has improved from the trough. We went into the pandemic growing our corporate business at 11% year-over-year. Growth fell to 9% year-over-year. Today, it's actually back to a positive, just under 1% year-over-year. We expect that number to continue to grow and improve, but do so slowly. While we believe that over a several-year period, we will get back to double-digit corporate growth, we think that path is going to take several years.
What's the, what's the mix there? I mean, I generally think of it as P times Q, right? volumes have been fairly soft recently. ARPUs are actually pretty good. part of the underlying trend in ARPU is shifting from a 100 meg to a one gig circuit, and ultimately, going to something higher than that. Maybe can you talk about that transition from a 100 to a gig to, I think, 10 gig?
I'm gonna answer your question talking about corporate and net-centric customers differently. For our corporate customers, they buy a fixed connection. It is an all-you-can-eat internet connection. The vast majority, well over 90% of our customers now, buy 1 Gb connections. two years ago, that was a 100 meg connection. ARPUs increased because of that transition to a one gig interface. We're actually continuing to see an improvement in the growth rate of our 10 gig corporate product. Our average corporate customer only uses about 5% of the bandwidth that they buy, so there's really not a P times Q calculation for those customers. They're buying a fixed connection. Why do they want the bigger connection?
There are three reasons: one, they want to have surge capacity; two, the differential in cost is relatively minor; and third, there is a, I think, false belief, but a belief in the marketplace, that greater capacity equates to greater quality. That's generally not true, unless you're approaching the limits of the port. Now, in pivoting over to our net-centric customers, the market is totally different. Those customers buy from us in data centers. They buy 10 gig, 100 gig, and 400 gig ports. They commit a portion of it. The company that was up here just prior to us entering the room, Fastly, is a large customer of ours, and we sell them literally hundreds of ports around the world to support their networks as we sell to their competitors. There, the customer pays on a usage basis, so P times Q does matter.
Traffic growth accelerated materially during the pandemic. It continues to be elevated. We continue to gain market share. Our rate of price decline has generally moderated somewhat due to the broadening of the customer base. Fifty-five percent of our net-centric business is outside of the U.S. It's spread among about 12,000 customers. There are some quarters in which the majority of the growth comes from a few large customers, typically hyperscalers, who get a lower price per megabit. That'll pull pricing down. There are other quarters where the majority of the growth comes from a broader set of midsize and smaller customers that results in a higher price per megabit. In general, the average price per megabit rate of decline has moderated somewhat over the past couple of years.
Going back to the corporate business a little bit, one of the things you've probably noticed within your business is differences in it from a geographic perspective in how the corporate business is doing. Are there markets where you're doing far better than that 1% corporate revenue growth rate? Then, what markets are they, and what type of percentage of your, you know, footprint do they make up?
You know, clearly markets like South Florida, Texas, even some other Sun Belt cities, Phoenix, Atlanta, doing extremely well.
Can you characterize, like, how they are growing relative to the total-
They're growing-
-historically?
They're growing as fast, if not faster, than our historical average.
Okay
Double-digit growth. We've got other markets that remain challenged.
Like double-digit declines, I would assume.
Some that are in decline. Bay Area, you yourself were a resident of Portland.
Yes
... for a period of time, and that was a market that's been very challenged...
Yes
... lately. Then there are some cities that fall right in the middle. You know, I would put New York in the middle of the pack, and that's actually a gorgeous market. Toronto, kind of in the middle. You've got some markets that are really outperforming and some that are clearly lagging.
... I have about six minutes left. We're gonna switch to the net-centric segment. If anybody has any questions, feel free to raise your hand. I'll try to get them addressed. On the net-centric segment, we've always talked about that business being driven off video traffic. I can't help but ask, with, sort of marquee sports properties going towards streaming, whether it was last year in Amazon and Thursday Night Football, this year in Google and NFL Sunday Ticket, what do you think that means to your business? Do you get more confidence when you see more and more sports properties go into streaming?
We carry traffic for both of those companies. You know, we clearly saw a shift in Thursday traffic with Amazon. We did have a relationship with DIRECTV for their streaming products, for the limited amount they did. You know, I think the YouTube relationship will be a bigger one. You know, in the pandemic, we saw a rapid acceleration in a trend that had been in place, that is the migration from linear video to streaming. Today, about 48% of all video consumed in OECD countries is streamed. That's up from 22% pre-pandemic. That's a much faster rate of market shift. These trends of sports migrating are usually at the forefront. You know, in Europe, we carry Premier League Soccer, Formula One racing, are examples that drive significant bandwidth demand.
Got it. Do you actually see noticeable shifts in traffic then, when some of those properties go online?
We do. You remember, our customer base, even in net-centric, is fairly broad. We have 7,800 access networks and nearly 5,000 content generators. We also see, for example, some of these larger streamers use sub-agents to help them. We will see an increase in traffic from the direct property owner. We will see third-party CDNs also pitching in to help that traffic distribution.
Love your perspective in terms of different applications besides video streaming that's gonna drive traffic. One notable one that I've covered since I cover Apple, is the Vision Pro, right? What's your view-
3,500, I'm not sure it's quite there. $3,500, I'm not sure it's mine.
Absolutely. Let's say they scale to several million units a year at some point in the future. What would your thoughts be in terms of, like, bandwidth consumption of that specific type of product?
Virtual reality and augmented reality are key drivers of the next leg in internet traffic growth. There will be some vendor and some product that succeeds. We carry some traffic with Apple. I think they've been aiming at a very specific demographic that is not the mass market. I think, you know, Facebook's transition to Meta and its product is more mass market, but maybe has not had the ergonomics or the content to be successful. I think ultimately, though, the next leg of growth will come from these higher bandwidth video experiences, and it's the combination of VR and AR, combined with unicasting. Remember, the history of content was designed around a broadcast world, and what the internet allows is for every stream to be unique to every end user, and that represents orders of magnitude, more bandwidth.
It also means that CDNs are less likely to distribute that, because it's inefficient to cache that many flows that are that different. There tends to be a little more centralization. We're seeing that with, you know, companies like Netflix, who have went from a more centralized model to a more distributed model, now back to kind of an intermediately distributed model.
Got it. We have about two minutes left. Are there any questions from the audience? Yes, George?
Can you imagine the cost of some DC to AC conversion? Yeah, like
the question was on cost for the DC to AC conversion in the data center facilities.
We have said that the total cost to connect the facilities and get them ready for sale will be about $50 million. We've actually already spent more than 50% of that CapEx in that project. The DC to AC conversion is actually relatively inexpensive to deploy those inverters. The big expense is in the batteries, the generators, the fire suppression systems that all exist and have been maintained in these facilities. The final piece of this equation is removing all of those literally tens of thousands of bays of dead telco equipment. Some of that is an expense on us. Some of that is actually an obligation of T-Mobile. It was equipment that they continued to hold title to. It was primarily wireless gear, and they are paying to have that equipment decommissioned and removed from those facilities.
That will actually be not an expense, but a revenue line item for Cogent.
Anybody else? With that, we're about out of time. Dave, thank you very much.
Hey, Brandon, thank you, and I'd like to thank every.