Welcome back to the AppsEconomy conference. For those of you I haven't had an opportunity to meet yet, I'm Mike Rollins, Head of the communication services and infrastructure stock for Citi. Just a few housekeeping items before we get started. I'd like to mention that we have disclosures available to the right of the video player as well as under the disclosures tab for Citi if you're viewing this via Velocity. Also, if you're joining us today and would like to ask a question during our session, you can enter your question into the question box and it'll come to us and we'll work to get that integrated into the discussion. With those housekeeping details out of the way, I'd like to welcome back Cogent CEO Dave Schaeffer and CFO Sean Wallace to our conference. Thank you for joining today.
Hey, Mike. Thanks for hosting us. As always, thanks, Citi, for a great venue. I wish I could be face-to-face with investors, but hopefully this virtual forum will be effective today. With that, Mike, I'm here to answer any questions you or the investors have.
Great. Well, thanks. As is tradition for this conference, Dave, we'd like to start off just learning about your strategic and operating priorities for the coming year. We'll start with that and then begin from there.
I am a person that kind of limits my all staff emails to the Cogent team. I actually only do one a year for the past 22 years on New Year's Eve, laying out goals for the next year. This year was no exception. Our goals are to increase our sales force size, increase the efficiency of our sales force, and to increase our footprint both in multi-tenant office buildings and carrier hotels. We have some very specific quantifiable goals for the team, and I'm a strong believer that the team can't achieve maximum results without having a clear objective in front of it. Now, clearly, all of this has been clouded and turned on end by COVID. The pandemic has lasted longer and been more severe, and the path out of the pandemic has been bumpier than we expected.
We're nearly 20 months into the pandemic, and if you had asked me back in March of 2020, we would definitely not be in this place today. You know, we sent our workforce home, the week of Christmas, going back to a remote work environment. Clearly the Omicron variant and the level of contagion has increased materially. You know, first and foremost, we place the health and safety of our workforce at the forefront. Now, one exception we've made is for our newest hires. One of the lessons we've learned from 20 months of the pandemic is that new hires do need in-person training.
For those employees that have less than one year of tenure in the sales organization, they are still required to come into a physical office and participate in face-to-face training in addition to online training. Our goals for the year are clearly to move beyond the pandemic, to resume our historic corporate growth rate, and continue to see our next generation growth and perform at historically high levels.
Great. Maybe just starting with the corporate side and, you know, what you were describing, anything new that you learned? Now, you said you allow your people home. What are you seeing from the buildings that you serve in terms of the employee bases, and has decision-making slowed down once again, or do you still see a recovering path in decision-making and sales opportunity in your corporate buildings?
Yes. We have a corporate on-net footprint of just under 1 billion sq ft of office space that is concentrated in large multi-tenant office buildings in the centers of major cities. We have the highest concentration in the Northeast, Chicago, Toronto, and Bay Area. Those markets have been the most impacted by COVID and have had the lowest occupancy rates in their buildings. To the South and in the Southwest, we have not seen the same kind of degradation. As we look at our footprint, we see that most of the lost customers is concentrated in between 300-400 buildings in those markets. We believe that we've lost and are through the worst of it.
You know, our corporate growth rate, which has historically been roughly 2.4% sequentially for the past 20 years, had a brief downturn during the Great Recession in 2008, 2009, but that downturn lasted only two quarters. The downturn of the pandemic has now lasted six quarters. The good news for us is that the past three quarters we've seen a negative growth rate at [STAR] each of those quarters. I think we are still a quarter or two away from returning to positive growth in the corporate business. From talking to our sales force and to customers, I think we're sensing from customers a level of fatigue. The management supported the IT departments to hurry up to get back to the office to get ready for a new hybrid work environment, only to then see those plans be pushed out again.
I think there are three major changes to corporate architectures that the pandemic has initiated. The first is larger connections at the primary location. A realization that high- speed symmetric internet connectivity is critical for a hybrid work environment. Secondly, I think companies have pruned their footprint, deciding which locations are no longer seminal to the organization and which locations will be needed for remote offices in the future, even if some employees are working from home. The third change is many companies have taken the lead from large companies like Citi or JP Morgan that have historically gone their dynamic ad hoc VPN concentrator in the data center. We're seeing a real uptick in corporate companies taking a small footprint in the data center as an on-ramp to their corporate networks. That presents Cogent an additional new corporate port opportunity with these customers.
I think as customers return to their offices, we've seen vacancy rates in cities such as New York over double during the pandemic, going from about 6% in New York City to about almost 17%. D.C. has gone from 8% to 17%, the highest in the city's history. San Francisco is nearly 30%. I think as employees start to get comfortable in returning to office, we'll see those vacancy rates go down. That presents an additional new opportunity for Cogent, because Cogent is most likely to win those customers when a new decision needs to be made.
Since 20%-30% of the office space will be making a new internet decision, that bodes very well for our corporate growth rate, re-accelerating and perhaps even passing the historical average for a period of time until we settle down into kind of a new norm.
In some of the markets that weren't as hit as difficult as, you know, in the Northeast, as we were getting examples of, are you actually seeing that rebound where maybe some vacancies picked up and you saw that kind of rubber band snap back and the sales got into substantially better proportion?
You know, we actually never saw the level of the trough as severe in those southern markets. While sales did slow somewhat, they did not ever go negative in those southern markets. Now they're kind of continuing at historical levels. Because our footprint is concentrated in very large buildings, the average building on the Cogent network is about 550,000 sq ft. I think, you know, most of those buildings are in the footprint that has been hit the hardest, and that's why our company-wide growth rate has turned negative for corporate growth.
In terms of the ARPU, you talked over the last year about how customers are adopting, you mentioned earlier, bigger connection and going for one gig. Is that still an opportunity just driving up ARPU for your customers? Well, maybe volume isn't as good right now. You're still seeing the benefits of a higher ARPU.
The answer is yes. Virtually all new customers are taking 1 Gb connections. One of the goals that I laid out this year was to increase the availability of 10 Gb connections in our corporate footprint to about 40% of the footprint. That does require some upgrades to our metro transport rings and in-building routers. Those routers are in a normal equipment refresh cycle. I think we will continue to see ARPU uplift in corporate on net from the high propensity of 1 gig sales and the. I think new customer realization at the incremental differential for a 10- gig corporate on net connection available is probably worth it. All in all, I think corporate on net ARPU are set to be stable to up going forward.
You mentioned earlier that one of the priorities for this year was to continue to increase the footprint. The corporate buildings that you serve, the carrier hotels or neutral data centers that you serve. Do you take that perspective or objective with what you were talking about in terms of the equipment refresh to enable 10 gig? Should investors expect something different on the CapEx line for Cogent?
No. Our CapEx efficiency should continue, historically as it has been. You know, as a percentage of revenue has consistently come down. In absolute terms, it's been relatively flat. Probably the biggest challenge for us on CapEx has been supplier availability of equipment. Normally, we would buy equipment with 30-day delivery terms. Now we're placing orders six-nine months out, with equipment sometimes coming in 30 days, sometimes coming in 12 months. You know, our vendors have experienced significant chip shortages. With certain equipment they've been unable to predict exact delivery dates. That has caused us to have CapEx be more volatile than we normally have it. It was up slightly in the third quarter. It probably will be above fourth quarter norms, but not because of the footprint expansion.
You know, as we look at our footprint expansion, our multi-tenant office building expansion will continue to be muted at 2.5%-3% of the footprint, 25-30 million sq ft. Since Cogent is really already in virtually all of the buildings that make sense, there are some new large buildings coming online. There are a few stragglers that we're adding, but I don't see an acceleration in our corporate on-net footprint growth rate. In our data center footprint growth rate, we for the last couple of years have been traversing about 100 carrier-neutral data centers a year. We have over 1,450 on-net.
As we look at the construction pipeline from our existing carrier-neutral relationships and new relationships, we think that number for 2022 will be somewhere between 100 and 125, which will be at that continued elevated rate. You know, the occupancy rate in many of these carrier-neutral is declining, but it is, I think, Cogent's top priority to be able to serve the widest carrier-neutral footprint of any provider in the world. It's why we have the largest footprint today as measured by third parties. Over 400 more data centers on-net than any other provider, and we have more networks directly connected to us, approximately 7,500 than any other provider by at least several thousand.
I think our strategy of being available at the core of the Internet globally has allowed Cogent's net centric business to grow at these unprecedented rates. As I mentioned earlier, with the need of many corporates to do that work from home intense concentration into the data center, I think having this big data center footprint will allow Cogent's corporate growth to also hit higher than historical average footprints.
As we move into the corporate customer, the net centric segment. I'll come to your comments there, Dave. Does that competitive position that you outlined give you a reinvigorated opportunity to become the number one share firm in that internet transit net centric arena? I think in the past you've referred to yourself as the number two player.
Yeah. I believe our market share is rising. Clearly, we would like to be number one. We have benefited from the fact that well over 70% of our traffic stays on net. That means seven out of 10, 7.5 out of 10 bits. Call a Cogent customer to a Cogent customer. While the ultimate goal would be to be 100% of the Internet, that's not realistic because then Cogent would be the entire Internet. I do think diversity makes sense for most customers. Our market share gain is continuing. Lumen has been the largest player, primarily through multiple acquisitions. They have de-emphasized transit as a product. We do not see them as a significant competitor in the market, and they are losing market share.
The number three player in the market, PEO, recently went through a divestiture from the parent company. There's been a lot of turnover in their organization and they are also migrating away from transit. With that, we think Cogent's market share will continue to increase. I surely hope by the time we're talking next year, I could report that we're number one as a market.
In this segment, you've had outsized growth over the past year. I think if my memory serves correct, it's been in the low 20s% year-over-year over the last nine months. Curious what you're seeing in terms of how sustainable that elevated growth can be for Cogent. You know, maybe an update in terms of whether it's the internationalization, whether it's acceleration of video, what you're seeing in terms of the drivers.
If I think about our net-centric business, it has always been a lumpy business with an average year-over-year growth rate of about 9%. The last three quarters had about 20% growth. It is the best three-quarter growth trend that we have had. There are really, I think, three drivers to drive that revenue growth. First is traffic growth, second is the location of that traffic, and the third is that traffic going to be two-sided or one-sided? The one thing that I did not mention specifically is the rate of decline in pricing. I know there's been a lot of talk in all industries around inflation. Technology is inherently deflationary. The Internet has been able to lower the price per bit at about 23% per year for the past 20 years.
We have followed that same price decline curve, and our margins have expanded while our competitors' margins have contracted because the architecture and platform that we have used is far more efficient than that of our competitors. As I look forward, I think Cogent will continue to drive down that 23% price decline curve, pricing its services generally at a 50% discount to our competitors, gaining market share and delivering outsized growth. While we may have several more quarters of this extraordinary growth, driven in large part by the acceleration in streaming because of the pandemic, over the long term, we think the net-centric growth rate at about 9% is the right thing for the company to anticipate and investors to model. In the short term, we may do better than that simply because we're gaining a disproportionate share.
More of that share is international and more of that is two-sided traffic. Our customer mix has skewed more towards mid-sized customers rather than all growth coming from the very largest customers.
Earlier, you were talking about the sales force, the importance of hiring and having that in-person training and development. The last quarter you were talking about some of that churn you experienced, in the sales force. You know, where are you in that process of stabilizing and growing the sales force? Has that elevated turnover impacted your ability to execute on fourth quarter sales?
First of all, our model has historically had high sales force turnover. Before the pandemic, we were averaging about 5.7% of the sales force turning over a month. When we went to remote, our turnover rate actually declined. It declined because it was harder for our managers to effectively manage and terminate underperforming reps. When we returned to the office, we implemented a mandatory vaccine policy coupled with mandatory return to office. Some employees did not agree with one or both of those programs. Secondly, we were able to identify the deficiencies in some of those moderately tenured reps and manage them out. Our turnover rate rose to 8.7%, so an over 50% increase from an already elevated level. I think our sales force turnover rate is going to come down because of the in-person training that we've put in place.
Omicron has been a challenge because while on one hand we want to keep our employees safe and allow them to work remotely in a highly infectious environment, we also have to balance that with the experience of low success rates with remote workers. This is a delicate balancing act. I think our sales force will grow. Our sales force productivity will increase. I think when we look at the fourth quarter, investors should expect trends very similar to that which they saw in the third quarter. I do feel that the NetCentric business is continuing to outperform. The corporate business is underperforming averages at a less dramatic rate, and our sales force turnover is declining as we have learned on how to manage and retain those employees in a hybrid environment.
Okay. The question I've asked you in the past is when you've gotten into situations where revenue has been a little more uncertain, you've taken some opportunities, a little more creative with the products to a very limited degree. A bunch of years back, you did a little bit more in data center. Recently, coincidentally or not, with the timing of the pandemic, you have a platform, an exchange platform that you've been creating, procuring for interconnection, in your own footprint and for your own geographic reach.
Can you talk a little bit about, you know, how you're thinking about trying to leverage your footprint in all of these, dense corporate buildings and carrier hotel data centers and, you know, if you're gonna look at that differently in the future in terms of the opportunity to introduce new products or, new services?
I take a fairly simplistic view of the business when I talk to the board and I say, we have this great asset. Our job is to maximize revenue out of this asset. I am not religious about products, services or technologies. I want to sell customers what they want on our footprint that will allow us to generate free cash flow from those sales. We have really expanded our transit portfolio in three distinct projects. We added VPN services back in 2006. Those remained relatively muted till 2014 when we were able to migrate from rapid spanning tree to a VPLS platform and have risen to be about 16% of revenues. Secondly, we have a number of data centers that are co-resident with our sales offices and the hubs of our network. They tend to be relatively small.
54 data centers, about 600,000 sq ft of raised floor space and about 70 MW of power. All of these data centers were taken over from failed operators. We did not construct any ourselves. We derive a little over 3% of our revenue from space and power sales in our data centers. We offer both our NetCentric and corporate customers a lower cost, easy on-ramp to the Cogent network. The third and final expansion was the establishment just about a year ago of an exchange platform allowing companies to not use transit, but to exchange IP traffic over our global footprint without a fixed commitment. That's a materially superior method of exchange that, when compared to the exchange points like DE-CIX, AMS-IX or LINX. It's often compared to some of the private exchange footprints, but our reach is more ubiquitous, lower cost.
We have seen traffic materially increase on that. It is still de minimis relative to our transit traffic and revenue. Roughly 81% of Cogent's revenues still come from selling Internet access. The reason we have been so focused on the Internet, it is where our competitive advantage lies. It is where the customer demand is. If you look at the marketplace of failed Internet providers and telecom providers, they all fought the inevitable trend of everything moving to the Internet. The Internet is more ubiquitous, lower cost, easier to use, and more application rich than any other platform. I think where we sit today, Cogent's future still lies with the Internet, but these supplemental products will help us drive incremental revenue on our network, giving it a place, it's a de minimis load on the network.
I feel very good about the footprint we have, the product set we have and the addressable market. What I don't feel good about is our corporate customers uncertainty around when and how they're going to return to work. From talking to these IT departments, I've come away with the understanding that they're just fatigued. They hurry up, deploy an architecture, get ready for back to office, only to see it delayed again. That's very frustrating for these companies.
One of the points you've made before is the breadth of vendors that you have from which you're leasing network elements to create the global Cogent experience. Within that diversity of vendors, are there any critical connections or elements that you have to manage through over the next two years, just given the consolidation that's taking, you know, place in the category?
We have over 285 diverse vendors. Our largest vendor by mileage is Lumen. We have nine years remaining on that agreement at no charge, and then we have negotiating rights going forward to extend. In every one of the links, there are alternatives. In fact, in some cases, we have activated alternate links because they were more optimal in terms of quality of fiber and the latency of the given route. I think we have no particular concerns of any vendor. Probably largest number of product orders is Zayo. A lot of small segments of metro markets, but they account for only a few percent of our total mileage, and those agreements run for another 18 or 19 years under fixed terms. In almost every instance, we have alternate solutions. We are in a golden age of fiber availability.
Even public companies such as AT&T and Verizon are now considering selling dark fiber, which would have been unthinkable earlier on as they are deploying capital to meet their 5G backhaul requirements. We have seen three major catalysts to fiber deployment. One being overbuilding metro markets to compete with cable. The second reason is consolidation of the world, drive speeds that was coming out of Lumen. All of those overbuilders make dark fiber available. The second catalyst has been the cellular backhaul builders. Those companies that justify their capital deployment solely on providing backhaul with significant surplus fiber. Crown Castle would be a great example in that market, Zayo to a lesser extent. And then third, there are a number of smaller business centric overbuilders that have emerged as well. Uniti and Windstream of the world.
You know, the Segra and the smaller regional players, FiberLight, here in the U.S. and dozens of these providers across Europe and even in Asia. I feel very comfortable that there is ample fiber available. Costs will actually come down due to the competitive nature of multiple providers competing on small footprints. With advances in wavelength division multiplexing, we have shown that with a very thin footprint, we can carry virtually all of the world's internet traffic.
One question on capital allocation, as you look out over time and you look at the trajectory that your dividend has been on, does the eventual migration to be a cash taxpayer change the scope of that line for Cogent? Where are you on that journey in terms of tax efficiency?
We have been, I think, an efficient allocator of capital, returning over a billion dollars to our shareholders, roughly $230 million through a buyback program that shrunk the float by about 20% and about $670 million through a consistently growing dividend that's grown for 37 consecutive quarters sequentially. Our dividend growth rate today is about 13%. It is well aligned with the growth in our cash flow. As we think about our tax policy, we have constantly looked at our non-U.S. non-expiring tax assets and have tried to use those more efficiently. While I don't think Cogent will be a zero cash taxpayer, I do believe that Cogent will be paying taxes at a substantially lower rate than the 21% corporate rate because of our ability to utilize those international NOLs.
No taxes in those regions and relatively low city and county taxes in the U.S. With nearly a trillion U.S. Dollars of non-expiring NOLs, we think we have a very long runway to being a low tax rate. While we have considered alternate corporate structures such as REIT conversions, we think that we should utilize our existing NOLs before taking any more aggressive tax strategy.
Going back to the question of, just cash allocation and dividends versus buybacks, you've kept a significant cash balance now for quite some time, you know, on the balance sheet. Any thoughts on how to, you know, maybe use that balance more effectively, or do you find it serves the company well to keep a larger cash balance just, you know, for whatever could come up in the future?
While I give myself a good grade in allocating capital for the company, I give myself a poor grade in being a macro market commentator. You know, I would have expected major market dislocations that would have allowed us to disgorge that cash quickly through a very effective buyback program. While markets have increased volatility, they continue to trade at historical levels. You know, whether it's the Nasdaq, S&P or Dow, all at all-time highs within a couple of % of those all-time highs. That is not an optimal environment to be using a buyback strategy. We think the dividend program is appropriate and is methodical. The cash balances are meant for those opportunistic periods to buy back stock. While I can't predict when a correction will come, I am absolutely confident that at some point it will.
The negative carry of waiting, i.e., carrying that cash balance with a relatively low cost of debt has been fairly denied. Again, I have two assets to optimize. One is the network, the other is the balance sheet of the company. The balance sheet of the company is strengthened by the underlying growth trajectory and operating leverage of the business. It's why I think we are in such an envious position when compared to virtually any other company in the telecom spectrum. Maybe I've been too patient in that allocation of excess capital, but it is the goal to eventually reduce those cash balances and get the share count to sufficiently reduce those cash balances by concentrating share counts.
Are you ready for our rapid fire three questions within three minutes?
Boy, as long-winded as I am, Mike, three minutes is tough for one question, but I'll try to get them. I'll be like I choke five to give you a one-word answer.
All right, first question. Why should investors buy your equity?
The FCF per share is dramatically underpowered.
Second question. Is inflation a net opportunity, net neutral or net risk for your business model and financial performance?
Probably a net neutral. We operate in a deflationary sector. That sector will continue to be deflationary, going forward. Even if there's general inflation, the price of Internet services are coming down.
Okay. Third question is, given that this is the Apps Economy Conference now, what applications can fundamentally shift these demands for connectivity and data consumption over the next few years? You know, I pose it, you have particular insight into, you know, some of these trends.
First of all, the Internet is as limited as the five million global users and their ability to publish applications. I have a great deal of faith in human ingenuity. I think we are on the cusp of moving into an augmented and virtual reality use of the Internet. The video experience will become much more immersive, much more interactive. Not possible in a unicast broadcasting world, but absolutely possible in a bi-directional Internet world. As fiber gets deployed closer to end users, I think we will see the explosion of augmented reality and virtual reality. Whether it's Metaverse or something else, you have to keep it fun. I think the user interfaces will get better, consumers will want it, they will spend more hours using it, and data consumption per unit of time will go up astronomically.
Okay, Sean, great to see you. Thank you for your time.
Hey, thanks, guys. Thanks everyone for staying up late. Take care.