Great. Let's get started. Good afternoon, everyone, and welcome to day two of Cowen's 50th Annual TMT Conference. My name is Greg Williams. I cover the cable, satellite, and telco sector here at Cowen. Today I'm joined by Founder and CEO of Cogent, Dave Schaeffer. We have a 30-minute fireside chat. I'll leave some time for questions in the audience. Without further ado, let's jump in. Dave, thanks for joining us.
Well, thank you, Greg, for hosting me. Thank Cowen for giving us the venue. Congratulations on 50 years, and maybe most importantly, thank the investors for taking time to hear a little bit about Cogent.
Sounds good. I want to jump right into the NetCentric business and particularly pricing. You know, just kicking things off. Cogent recently reported another quarter of strong growth with revenues of 18.6% year-over-year ex-FX in the quarter and it was driven by healthier pricing trends. Can you break down the individual components of the better pricing Cogent receiving? I know you mentioned, like, a broader set of customers, more European customers, and are these pricing trends sustainable?
Yeah. Forty-two percent of Cogent's revenue, but 95% of Cogent's traffic comes from our NetCentric business. This is where we're selling bulk internet connectivity on a metered basis in data centers in 51 countries around the world, 1,450 carrier neutral data centers. Within those data centers, we have two customer bases. We have companies that pull down content to distribute to their end users. We connect and sell transit to over 7,600 access networks around the world. Those networks actually operate in over 170 countries, so in some cases they come to a data center to meet us. We sell to about 5,000 content generating businesses. The average price per megabit on the internet has declined at a compounded rate of approximately 23% per year for the past 25 years.
That trend is going to continue. The total addressable market for the sale of transit globally is about $1.5 billion. Cogent has averaged about 9% revenue growth by gaining market share in that transit space. Now, while the nominal price continued to fall at pretty much historic rates, actually 22% on a year-over-year basis, our effective rate of price decline was much more moderate, and there are three contributing factors to that. The first one is that we have been selling more to smaller customers, and smaller customers tend to pay a higher price per megabit. The second reason our effective pricing has been better is that much more of our traffic is traffic that we're getting paid by both sides.
When we connect to other networks, because the Internet is not a singular network, but an amalgamation of networks, there are two ways that traffic can be exchanged. One is to a transit provider such as Cogent, the other is through a peer. We have a limited number of peers. We have 24 peers, and those 24 peers account for approximately 27% of exchanged traffic. In that 27% of traffic, we get paid by our customer, but we receive no revenues from our peer, and we pay nothing to the peer. In 73% of the cases, the traffic is going from one Cogent customer to another. In those cases, we're getting paid by both sides. That percentage has increased over the past several years from roughly 50% to 73%.
That has had a meaningful effect on us getting a higher price per megabit of revenue, even though the nominal price continues to fall at the same headline rate. The final reason is that the Internet is becoming much more globalized. 22 years ago, 85% of the Internet originated or terminated in the U.S.. U.S. is only 4.5% of global population. Eventually, we should only be 4.5% of global traffic. Today, the U.S. is about 1/3, roughly 33% of global traffic. Cogent is particularly strong internationally, where we serve customers on every continent, and we are more likely to have both the content and access networks as customers.
Most of our peers are U.S.-based, so because the rest of the world's traffic is growing faster than the U.S. traffic, that has helped our effective price per megabit as well. The final point I'd like to make is, while we had the second best quarter in the company's history, and actually the quarter before was the best quarter in our history, in terms of NetCentric revenue growth on a year-over-year basis, our long-term average growth rate is about 9% per year. I think over the next year or two, we will revert back to those more long-term averages. Just to remind investors, heading into the pandemic, we were actually growing the NetCentric business below trend line at only about 4%. It's really been the acceleration of streaming and the migration of linear video to over-the-top that has accelerated the growth in international traffic.
Got it. Smaller customers, more two-sided payments, and the internationalization of the Internet are driving your pricing power. Maybe on point number two, the two-sided payments. You said it went from 50% to 73%. You know, where could that go at the end of the day? Because you've got the pricing power and all the data centers, and how much more share can you take? Is this sort of plateauing here?
We are connected to more carrier-neutral data centers as measured by third parties than any other provider in the world by almost 400. We're in 1,450. Our next closest competitor is in roughly 1,000 data centers. These are not central offices, but true carrier-neutral data centers. We are the lowest-cost provider. We guarantee to undercut our competitors by 50% and continue to gain market share. Now, we're greedy. We would like to have 100% of the traffic being two-sided, but I think that's unrealistic because that would mean the entire Internet globally rides solely on Cogent. I think most of our customers look for second and even tertiary providers for diversity. I do think we can get that 73% up. It'll probably plateau high 70s, low 80s, in terms of two-sided traffic.
Today, we are the second largest carrier of Internet traffic in the world. We carry about 25% of global bits. That gap between us and number one is shrinking, and I do think that our continued gain of market share will have Cogent as the largest provider for the next several years. The final point I'd like to make is we are more connected than any other network in the world. The Internet is comprised of about 60,000 autonomous system numbers. That's the way networks talk to one another. We have 7,625 of those as direct customers. That's almost a 50% greater number of direct connections than our next closest competitor.
Got it. Let's just talk about traffic a little bit. Starting in the second half of last year and in the first half of this year, traffic's decelerated from elevated levels, you know, from the pandemic, to the mid to high teens year-over-year. Do you anticipate traffic growth continuing to decelerate perhaps closer to 10% or even below? Or do you think it could stay at this, you know, 17% year-over-year level?
The long-term growth in traffic over the past 25 years has actually been about 24% per year. We're below that. I think we had a meaningful acceleration at the beginning of the pandemic. Now, as much of the developed world is emerging from the pandemic, lockdowns, we're seeing traffic growth revert back to more normalized patterns. You know, I know that some of our CDN customers have publicly commented on slowdown in growth. We're not seeing that. I think our traffic growth will be comparable over the next several quarters.
Traffic growth is comparable, but you're saying long term, NetCentric will revert back to this 9% revenue growth. Help us with the algebra there. You know, the smaller customers are gonna be fewer and far between , or, you know, the internationalization of Internet is gonna Dave Schaeffer, just help me out with how you get to 9% moderation.
Yeah. I think it'll be a combination of those factors. I think, one, eventually, you know, the Internet will reflect the population base of the world. Two, we will see the customer bases normalize. You know, we went from a period where three or four bang names were dominating. While they're still very material, they are not necessarily as dominant as they were. You know, my guess is we'll see some customers win and some lose in the market. And then that migration and two-sidedness will continue but probably moderate. As I put all these factors together, I don't have perfect visibility to revenue growth even day to day. You know, this is a usage-based service. Over the long run, we definitely see a trend back to kind of the long-term average. It's the best metric I can give.
Got it. Just switching over to Corporate on-net, you know, NetCentric outperformed, but corporate's still trying to find its footing. You know, positively corporate on-net declined only 0.8% quarter-over-quarter. It's the lowest Q-over-Q decline since the start of the pandemic, I believe. You noted in the earnings call that Cogent's leading indicators are trending positive. It's like badges going to the buildings and real estate tours, et cetera. Have these trends continued into the second quarter here thus far, you know, as we think about maybe a slower return to office , especially in areas like California and Northeast?
Yeah, I think we are seeing a continued return to office and a solidification of a hybrid work model for most companies. You know, the pandemic was very disruptive. Our corporate business prior to the pandemic had grown 51 out of 53 quarters sequentially. There were two negative quarters, that being Q4 of 2008 and Q1 of 2009 in the financial crisis. Fast-forward to 2020, and we've now had eight consecutive quarters of negative growth against the business that would normally deliver roughly 2.5% sequential growth. We are seeing companies make their buy decisions. We're seeing a number of leading indicators, both directly and indirectly, and let me touch on those. Directly, we're seeing our sales force activity improve, the number of proposals that are issued increase, and the gestation time of those proposals from issuance to order shrinking.
Companies are ready to move forward. They had two false starts gearing up to return to office, only to be shut down with Delta and then with Omicron. Now companies are pretty much, "We know where we're going. Yes, there may be additional pandemic-related issues, but we got to get on. We can't let our networks sit two years in the past." The second thing that's happened to us is in our corporate buildings, we have roughly 1 billion sq ft, about 1,850 skyscrapers. The vacancy rate in those buildings has gone up from 6% to 16%. There's just less businesses in the buildings. Now, what we are encouraged by is an external metric, which is number of employee entrances in the building, i.e. security badge swipes. There, there's a fair amount of geographic dispersion.
In the best markets were between 60% and 70% of pre-pandemic levels. The worst markets were in the mid-20s. New York, somewhere in between. San Francisco is probably the worst of all markets. We're seeing companies adopt a permanent hybrid work model, and that has, I think, three consequences. One, it means they need a bigger connection at their primary location to support those ad hoc VPN sessions. Two, they want to establish a second failover point for that VPN concentration function. Third, for those companies, they'll probably take slightly smaller office space. Now, we sell our corporate service not on a meter basis, but on a fixed connection. If each tenant in the building takes less space, and the building returns to its historic occupancy rate, we'll actually have a bigger total addressable market.
More customers per sq ft.
More customers in the building. That's exactly right. More customers per square foot. That's a positive for Cogent. Going into the pandemic, our average building had 51 businesses in it, and the building was about 550,000 sq ft, so about 11,000 feet per tenant. If that shrinks by 20% to 9,000 feet, the building tenant count goes up to over 60, giving Cogent a bigger market opportunity.
Right. They're taking bigger ports. They might be taking a second or even third port. Like I said, we're seeing more customers per building. Maybe, on that first point, you know, it's been another positive as customers migrate from 1 Gb per second products to 10 Gb per second products. What percentage of the customer base currently uses the 10 gig product? Do you think Cogent's entire customer on-net base will ultimately go there?
I'll take those in reverse order. I think the answer is ultimately yes. I remember when I originally went out to raise private capital for Cogent, I had investors laughing at me, telling me no business would ever need 100 Mb of internet connectivity in their office. It was just overkill. This is when a big connection was a T1 at 1.54 Mb, and many customers were still using 256 Kb DSL connections. Now, the applications have become much more robust, more bandwidth hungry, and businesses become much more dependent on their internet connection. We saw a migration away from 100 Mb to Gb connections, where today over 90% of our base is taking our 1 Gb interface. 10 Gb still a relatively de minimis part of the base. It's sub 1%.
It's not available in every corporate building, and it is also way more than a customer would need. Our average corporate customer on a 100 meg connections was using about 18%. As they migrated to 1 Gb connections, our average corporate customer utilization declined to 11%. The port got 10 times bigger, and they lost less than 50% of their peak utilization. It did say that they got benefit from that 1 gig migration. I think it'll probably be several years before the customer premise equipment that supports 10 gig is prevalent enough and the applications become bandwidth intensive enough to need that kind of capacity.
Right. We're seeing the return to office slowly come back and maybe the solidification of this hybrid workforce environment. Things are going in the right direction there. Unfortunately, looking at recessionary possible pressures, what does that mean for your business as we think about your sequential growth, and when you can achieve that in corporate and at a sustainable level? How are you thinking about it?
Yeah. Cogent's lived through, you know, the dot-com recession, the financial crisis, and then the pandemic recession. I feel that our service is relatively recession resistant for two reasons. One, it is a necessary utility, so companies can't do without the Internet. They can do without many other services. Telephones, for example. If you surveyed a business 20 years ago and said, "If you have to turn one of these two utilities, a phone or the internet off, which one would you turn off?" They would all say internet. Today, if given that same choice, I think universally they would all say, "The internet can't go away. We'll turn even, you know, heat off before we turn the internet off.
Sure.
Those are important benefits. The second thing that's helped us is the footprint that our corporate business originates from. We go into the most expensive real estate in a market, big, tall, shiny skyscrapers. The landlords have done a very good job in vetting those tenants. They have a lot more at risk. Our bad debt is some of the lowest in the industry. It actually declined during the pandemic to about 0.8%. Our DSOs declined, again, probably the best in the industry, hovering around 21-22 days. That's telling us our customers are paying their bills. Our churn rate did not go up. I think that a general recession will not have any material impact on our corporate business. I think over the next several quarters, that business should be able to get back to its long-term growth trajectory.
Interesting. Maybe we can feed off of that last statement. When do you think then? What's a reasonable timeline for returning to Cogent's historical 2% sequential growth?
You know, I think it's a couple of quarters away, whether it's two or three. We don't give specific quarterly guidance, but we're seeing significant leading indicators, and we saw a significant inflection from Q4 to Q1, where that corporate growth rate went from -2.5% to essentially flat, down 0.1% on a sequential basis. I think, you know, we'll probably see a slower rate of increase coming out, but I think we're two to three quarters away.
Got it. It's helpful. Just wanted to flip to the balance sheet really quickly. Your leverage remained around 3.58 times in the first quarter at slightly above the high end. You typically are at 2.5-3.5 times. We talked about maybe taking leverage up, as you think about the dividend, et cetera. Are rising rates changing that thought process?
We view our balance sheet as a strategic asset. We use leverage to accelerate returns of capital to our shareholders, not to build our business. We built the business completely leverage free. Even though rates are going up, we're still in a historically low interest rate environment. We have two tranches of debt outstanding. We have our $500 million U.S. denominated senior secured debt at 3.5%. We actually lowered that interest rate even more by doing a fixed to floating swap, bringing our effective interest rate in at around 2%. In addition to that, we have our euro-denominated unsecured debt that has two years of maturity.
We will most likely take that note out and replace it with a U.S.-denominated issuance that'll allow us to crystallize the currency gain that we have achieved since issuing that note, probably monetizing about $30 million in non-cash gain becoming a cash gain. We will also realize that there will be higher cost of capital. Probably we'll stick with a shorter duration. Now to our total leverage. We're using the capital primarily to help us accelerate our dividend growth as well as occasionally buy back stock. We have 39 sequential consecutive quarters of raising our dividend. You know, while I can't give an exact prediction in the future, past is a pretty good indicator of what we're gonna be able to continue to do. We kind of view our balance sheet as underlevered. We have top line revenue growth, albeit slower during the pandemic than pre-pandemic.
Our average growth rate, absent the pandemic, had been around 10% with 200 basis points of margin expansion. Pandemic hits, even though NetCentric has outperformed, corporate underperformed, total growth rate decelerates to about 5%. Margin expansion decelerates to about 100 basis points. That's for the past two years. Much better than any other wireline company, but not great by Cogent long-term averages. As we start to come out of the lockdown phase of the pandemic, we've seen, you know, our corporate business rebound, and we should get back to that long-term average of 10% growth and 200 basis points. With that, there's a lot of natural de-levering, and we view the capital that we have as kind of a toggle between either a dividend recap or a creeping LBO. We've done both buybacks and dividends.
We've erred more over the past few years on dividends because they've been tax efficient. Last year, we treated 79% of our dividend as return of capital. Our view is leverage is going to hover in this range, maybe drift down a little, but we have plenty of capacity as the EBITDA continues to grow.
Got it. I wanted to talk about margins as well. You touched on it by saying that there's been margin pressure through the pandemic, but now we're coming out of it. Of course, we've got inflation related pressures now. You're expanding your sales force. Your product mix is shifting. You're diluting the margin because you're going to more corporate. How should we think about the trajectory for EBITDA margins over the course of the year?
I would love to sit on this stage and tell everyone we have pricing power and price of Internet's going up. That's a fantasy. That's not real. The price of Internet traffic is gonna come down. The technologies that allow us to produce that traffic are getting more efficient. We produce interface-routed bit miles that create Internet traffic. We think that headline deflation number is gonna remain in the low twenties, just as it historically has, because there's no real change to the advance rate in technologies. Now, yes, some of our ancillary costs will go up, power will go up, some of our employee costs will go up, but we can offset those through better efficiency. So for example, on a kilowatt per bit mile, we've had a 16.5% compounded improvement per year over the past five years.
That's how long we've been measuring it. That's both ESG friendly plus financially friendly. You know, in terms of product mix, yes, we'll sell a little more corporate, it will be a little more off net. When we were hitting 10% growth and we had that same mix, we were delivering 200 basis points. We feel pretty comfortable that our long-term multi-year guidance is correct until our EBITDA margins ultimately plateau at around 50%. To remind investors, are around 39% today.
Got it. You're talking about efficiencies, but at the same time, you are hiring in droves. I think you're targeting 7%-10% sales growth annually. Is this target realistic, given, you know, the great resignation?
The answer is it's absolutely realistic on the hiring front. I'm not as certain on the retention side. We went into the pandemic growing 7%-10% for the previous three years, and we continued to hire. We actually saw our hiring rate accelerate, but our turnover rate accelerated even more. We pivoted our training to remote training, but it was not nearly as effective as the training we did in our offices. As we return all Cogent employees in March to offices full time, we think that should inflect our turnover rate down. Our turnover rate in sales has always been high. It's about 5.2% of the sales force a month on average. During the pandemic, it peaked at 8.9%. It drifted down to 6.9% in this last quarter, but still above the 5.2.
This is because the people that we hired remotely were never able to build a robust enough funnel to hit their success targets. I think as we now have people back in the office and we can mentor them and give them additional training, we'll see rep productivity go up and rep turnover rate decline.
Got it. With that, we're out of time, Dave. Thank you very much.
Thank you very much, Greg. Welcome aboard on covering us.
Thank you.
Welcome back, I should say.
Yeah, welcome back.