Good morning, and welcome to the Cogent Communications Holdings Third Quarter 2021 Earnings Conference Call. As a reminder, this conference call is being recorded, and it will be available for replay at www.cogentco.com. A transcript of this conference call will be posted on the same website when it becomes available. Cogent's summary of financial and operational results attached to its press release can be downloaded from the Cogent website. I would now like to turn the call over to Mr. Dave Schaeffer, Chairman and Chief Executive Officer of Cogent Communications Holdings.

Hey, thank you, and good morning, everyone. Welcome to our third quarter 2021 earnings conference call. I'm Dave Schaeffer, Cogent's CEO. And with me on this morning's call is Sean Wallace, our Chief Financial Officer. Now for a few comments on our results. As the focus of the pandemic-related efforts of companies have shifted to a broad reopening of the U.S. and global economy and as most large businesses have developed plans and deadlines to reopen their offices, we've seen signs of improvement in our corporate business climate. However, despite this improvement, the delta variant has delayed a large portion of these back-to-work plans to early next year and many key indicators of office activity remains significantly below normal levels. And many U.S. northern cities and in Canada, we see that rates of employees working in offices and the leasing of commercial office space and central business districts remain significantly below historical levels. Our NetCentric business continues to benefit from the greater-than-expected growth in streaming subscribers and the continued internationalization of the Internet, where our global footprint positions Cogent as the best network to deliver end-to-end on a global basis for our customers. For the third quarter, our traffic growth moderated somewhat from the fast pace in the previous periods, but was up 1% sequentially and 25% on a year-over-year basis. Despite these improvements, we remain cautious in our near-term outlook on the uncertain economic environment and the challenges that are continuing because of the pandemic. On a U.S. GAAP basis, our revenue was up slightly to $147.9 million in the quarter, an increase by 4% on a year-over-year basis. On a constant currency basis, we experienced sequential quarterly revenue growth of 0.5% and an improvement in our year-over-year growth rate to 3.6% from the 2.8% constant currency growth rate…

Thank you, Dave, and good morning, everyone. This earnings conference call includes forward-looking statements. These forward-looking statements are based upon our current intent, belief and expectations. These forward-looking statements and all other statements that may be made on this call that are not historical facts are subject to a number of risks and uncertainties, and actual results may differ materially. Please refer to our SEC filings for more information on the factors that could cause actual results to differ. Cogent undertakes no obligation to update or revise our forward-looking statements. If we use non-GAAP financial measures during this call, you will find these reconciled to the GAAP measurements in our earnings release, which is posted on our website at www.cogentco.com. A quick update on COVID-19. Like many companies, Cogent continues to be impacted by the lingering effects of the COVID-19 pandemic and the accompanying responses by governments around the world. In October of this year, our entire U.S. workforce returned to our offices after 18 months of working remotely. A majority of our offices in the rest of the world continue to work remotely, although our employees outside of the United States are scheduled to return to their Cogent offices by the end of the year. I want to thank the entire Cogent workforce and in particular, our IT and finance departments for their continued hard work during these very challenging times. I also want to thank our field engineers, contractors, billing and collection staff and many other Cogent employees who continue to work on the front lines, installing our new customers, maintaining and upgrading our network and providing outstanding customer service. The COVID-19 risks and other risks are described in more detail in our annual report on Form 10-K for 2020 and in our quarterly reports on Form 10-Q for the quarters ended September 30, 2021, June 30, 2021 and March 31, 2021. Throughout this discussion, we will highlight several operational statistics. I will review in greater detail certain operational highlights and trends. Following our remarks, we'll open up the call for questions and answers. Now I'd like to turn it back over to Dave.

Thanks, Sean. Hopefully, you had a chance to review our earnings press release. Our press release includes a number of historical quarterly metrics that we report on a consistent basis. Our targeted long-term EBITDA margin expansion guidance is for an annual improvement of 200 basis points. Our targeted multiyear constant currency growth rate targets remain approximately 10%. Our revenue and EBITDA guidance are intended to be multiyear goals and are not meant to be used as either quarterly or specific annual guidance targets. Our corporate business, which represents 60.2% of our revenues, our corporate business declined by 1.5% from the second quarter of 2021 and a decline of 6.9% from the third quarter of 2020, primarily due to the impact in the pandemic and some reductions in ARPU due to lower loop costs for our off-net customers. Our NetCentric business, which represents 39.8% of revenues, had another strong quarter, and we achieved a continued growth rate of 2.5% quarter-over-quarter and grew on a year-over-year basis by 26.3% from the third quarter of 2020. Volatility in foreign exchange rates primarily impacts our NetCentric business and slightly over 50% of that NetCentric business is outside of the United States. On a constant currency basis, our NetCentric business increased by 25.1% in the third quarter from the third quarter of 2020 and 3.8% from Q2 of 2021. Now Sean will give some additional detail on our financial performance for the quarter.

Thanks, Dave, and again, good morning to everyone. Let's talk about corporate and NetCentric revenue and customer connections. We analyze our revenues based upon network type, on-net, off-net and non-core, and we also analyze our revenues based upon customer type. We classify all of our customers into 2 types, NetCentric customers and corporate customers. These customers are typically professional service firms, financial service firms and educational institutions located in multi-tenant office buildings or connecting to our network through our CNDC footprint. Our NetCentric customers buy significant amounts of bandwidth from us in carrier-neutral data centers and include streaming companies and content distribution service providers as well as access networks who serve the consumers of content via fixed line and mobile networks. Revenue and customer concentrations by customer type. Revenue from our corporate customers for the quarter declined sequentially by 1.5% to $89.1 million and declined year-over-year by 6.9%. Corporate revenues in the quarter, excluding the impact of USF taxes declined by $1.4 million. This is the fourth quarter of consecutive smaller declines in our corporate business, which peaked at a decline of $2.2 million in the fourth quarter of 2020. An increase in the USF tax rate which only applies to our corporate VPN connections had no material impact on a sequential basis and had a $0.9 million positive impact on our year-over-year quarterly corporate revenues. The USF tax rate changes quarterly, and we cannot predict the impact of future USF rate changes on our revenues. While the delta variant slowed down the level of corporate customer activity, we continue to see indications that our corporate business is trending towards pre-pandemic levels but at a slower pace than expected. We are encouraged by the continued decline in the churn of our 100 megabits per second and 1 gigabit per second…

Thanks, Sean. I'd like to highlight a few of the operational strengths in our business, our network, our customer base and our sales force. NetCentric performance. As I stated earlier, we continue to see strength in our NetCentric business as revenues increased by 26.3% year-over-year. Streaming service providers are aggressively targeting overseas markets, and we are a direct beneficiary of this growth. We have positioned our network on our capabilities to support the growth in streaming on a global basis, and I'd like to highlight some of these important characteristics. At quarter's end, we connected to 1,332 carrier-neutral data centers as well as 54 Cogent-operated data centers. We connect to more data centers than any other carrier globally as mentioned by independent third parties. The breadth of this coverage enables our NetCentric customers to better optimize our networks and reduce latency. We expect that we will widen our lead in this market as we are planning to add approximately an additional 100 carrier-neutral data centers per year to our network each year over the next several years. At quarter's end, we directly connected to 7,590 networks. This was an increase of 5.2% from a year earlier. This collection of ISPs, telephone companies, cable networks, mobile telephone operators and other carriers allow us to provide access to a significant majority of the world's broadband and mobile phone users. This critical mass of eyeballs makes us an extremely attractive service provider to streaming providers looking to directly connect their customers improving video quality and download speeds. At quarter's end, we had a sales force of 218 professionals focused primarily on the NetCentric market. We believe this group of professionals is the largest and most sophisticated of such sales teams in the industries. Now for some of our corporate trends. We are seeing…

[Operator Instructions] Our first question comes from the line of Sami Badri with Credit Suisse.

It's George on for Sami. I have 2 questions. So can you provide any additional color on return to office timelines for your customers and how that may impact the timing of Cogent's return to sequential growth in the corporate segment, obviously, in light of the slower-than-anticipated central business district leasing? And then my second question is in regards to the ongoing labor shortage and how, if at all, that may impact your previously stated sales force growth aspirations given the large quarterly headcount decline?

Yes, sure, George. So with regard to return to office, a number of commercial real estate firms surveyed the market consistently. They provide third-party data around 15 key KPIs measuring the health of the commercial office market. Probably the 2 that are most relevant to Cogent are badge entries or security entries per day and the pace of leasing activity, both primary and subleasing. What we have seen is that in the northern cities, we're still running at around 35% employee entrances per day versus pre-pandemic levels in Q1 of 2019. In the southern cities, those numbers are up to between 60% and 70% of badge entries per day versus pre-pandemic levels. Those numbers are continuing to increase, and we think this is maybe the most important indicator. It's also important to remember that we sell our corporate product on an unmetered basis. So whether there's 10 employees in their office or 100, the customer pays us the same. And the majority of our corporate customers have historically had VPNs that are aggregated through their firewall in their corporate office. A positive trend that we have begun to notice as companies define a hybrid work schedule for their employees. We have seen some corporate customers taking additional port from Cogent in a carrier-neutral data center solely for that VPN aggregation function. This should present an additional opportunity for us to sell more ports to those corporate customers, offsetting the decline in office-to-office VPNs as companies reduce the number of offices. When we put these various puts and takes together, we believe that our corporate business is still probably several quarters away from returning to its historic sequential growth rate. Again, to remind investors, this is a business over 16 years that averaged better than 2% sequential growth. The only other…

Our next question comes from the line of Phil Cusick with JPMorgan.

So can we dig into the corporate gross adds funnel sort of customers picking up the phone? And as you think about your headcount turnover, should we think of smaller sales forces slowing sales for a decent period here as those new reps train up? Or is that not an issue?

Phil, thanks for the questions. I'll take those in reverse order. The reps that we have been managing out have been the less productive reps. Actually, our average rep tenure at Cogent had the largest increase sequentially quarter-over-quarter, going from about 31 months to 33 months. So we've been able to retain good reps and our model, while it sounds easy to just pick up the phone and sell Internet is very hard, and it requires a certain amount of organizational and discipline among our applicants. We've always experienced high sales force turnover at over 5% of the base per month. The elevated turnover at [indiscernible] , I think, is a transitory event due to the mandatory vaccinations and the return-to-office requirement, while some employees had acknowledged they were willing to do so when they will require to chose not to. And then we became much more disciplined about looking at underperformance. We realized the status of the underlying market. And for that reason, our corporate sales force declined by about 9%. It just did not pay to carry marginal reps. We, however, do expect to continue to grow the sales force coming out of the pandemic. With regard to the NetCentric sales force, the decline was much smaller at about 5%. And again, many of the same reasons, but much less underperformance and quite honestly, a healthier underlying market demand. Now I'm going to pivot to your first question, which is what we're seeing from corporate customers. And I have to admit, as a CEO of an operating business, I believe the impact of the pandemic was going to both be shallower and shorter than what turned out. We saw virtually all of our customers mirror our behavior on remote work. And they initially had anticipated return to office…

And I'll just add some color to Dave's comments. If we look at the 3 quarters before the pandemic and then 3 quarters of this year, quarter 1, 2, 3, and look at the components, gross adds, churn and ARPU, ARPU has been right where we thought it would be. It's been very consistent. Churn has actually improved and is actually -- we're at better levels than where we were pre-pandemic. And as Dave has mentioned, because of deferred openings of offices and deferred filling of vacancies, our gross adds are still below where we were on the pre-pandemic basis. So we're optimistic that as the world opens up and offices -- corporates go back into offices and begin to reconfigure their networks and as new tenants go into buildings, we'll get back to a more historical level.

Our next question comes from Colby Synesael with Cowen.

Great. Two modeling questions, if I may. One on cash flow from operations CFO. You flagged that it was particularly strong in the quarter. Anything worth flagging and whether or not we should see a little step down in the fourth quarter? And then secondly, I guess, similarly on CapEx, you mentioned you pulled forward some purchases. Should we see a corresponding reduction in the fourth quarter? Should fourth quarter look like a more normalized level? And if fourth quarter looks like a more normalized level, should we expect then a more notable reduction as we go into 2022?

So let me take those on. The cash flow, I think we expect it to be consistent, although the DSOs at 21 days are the best in the company's history. We see no indication that's going to deteriorate, but that does have an impact on our cash flow statistics that we report. But we do expect kind of a similar type of elevated level of cash flow from operations for the fourth quarter. In terms of equipment purchases, I'll be honest, I don't have perfect visibility to that answer. And the reason is as the chip shortage has continued and I think morphed and become more widespread, we are dependent on a single vendor, Cisco. We have daily conversations with them about equipment availability. Typically, we would be able to order equipment and expect delivery within a month. Now we're going to 6-month forecast for delivery with virtually all products experiencing daily volatility and expected ship date. So it's very common for the items that we have on order to get daily reports saying they're being pulled in and then they're being pushed out. We tried to get ahead of this. We did pull in several million dollars of orders. We did pull additional orders in. But I can't tell you when that equipment is actually going to arrive in our warehouses based on the inability of our vendor to do that. When we look at the base level of capital, it's about flat with where we were last year, but the elevated level is due to these accelerated orders now rather than having orders pending for 30 days. Cogent has 6 months of orders outstanding with its vendor and could they show up early and therefore, elevate CapEx in the fourth quarter? Possibly. Could they be delayed even further than the 6 months, and therefore, CapEx will be substantially below. I just don't want to represent to investors something that's out of our control.

So Dave, I guess, just to be clear, you not only made orders, but you actually received in your warehouse in the third quarter, some of those shipments that presumably were made, I guess, a quarter or 2 ago that resulted in the higher CapEx this specific quarter. And I guess, to your point, you've also made incremental elevated orders, which could result potentially in elevated CapEx, whether it's in the fourth quarter, some future quarter, you're not sure. Is that -- am I hearing that correctly?

You are hearing that absolutely correctly, Colby, because we record the CapEx upon receipt of the equipment.

Got it. And then just one quick one, if I may. On leverage, you're not at the high end of your target, 2.5 to 3.5. Could that potentially have any limiting factor on dividend growth going forward?

We do not anticipate that. We also note that about 6 or 7 basis points of that is due to foreign exchange distortion. So under the indenture at [ 3, 4, 7 ], if we took the FX out, we'd be at about [ 3, 4 ]. Yes, that's above the midpoint, but still below the high end. That high end is an internally set goal by Cogent. It's something that's reviewed with the Board as we did a sensitivity analysis for the Board earlier this week. We feel very comfortable that we're going to be in a position to continue to grow our dividends for the foreseeable future.

Our next question comes from Nick Del Deo with MoffettNathanson.

First, since the return to office trend has been a bit different in, you call it, coastal cities or northern cities versus some in the South, are there trends you can observe for metros with higher in-person office attendance versus those with lower rates? Yes, that might give us a sense for how the corporate business should recover as we start to see the cities that are behind start to look more like the cities that are further along.

Yes, sure, Nick. Great question. So our sales force sells nationally. So we can't look where the sale originated because a salesperson in New York may actually sell a customer in Phoenix or a salesperson in Houston may sell a customer in Seattle. But we can look at where the orders are installed, and we have seen a better performance, lower churn and higher installs in those southern cities. Is it enough to really give us confidence that, that's where New York and Toronto and Chicago were going? I'm not sure. These are smaller markets. They tend to have a different type of business. So for example, if we look at our footprint in the Texas market, while it's a broadly diverse economy, it's much more heavily dependent on energy. New York is heavily dependent on financial services. That tends to be a positive for us because financial services tends to be an industry vertical that is more prone to aggressive return to office than some of the consulting and service businesses. Legal, however, tends to be an industry that's a laggard in return to office. So I think we look at the data both by endpoint geography and by SIC code of the customer. When we put all of that data together, I think we're pretty encouraged that as badge swipes increase and leasing activity increase our corporate sales rate in those buildings will increase.

Okay. Okay. That's helpful. And then a couple of questions or clarifications on the sales force front. First, it looked like your full-time equivalent reps went up in the quarter despite the total number going down quite a bit. Just mathematically, how does that work? And then second -- sorry, go ahead.

I'm laughing. I actually -- I went to [ Kesha ], who does our financial modeling and asked the same question. And the reason is the FTE is a weighted number over the month and the number of reps reported is an ending number. So we had a significant amount of terms on the last couple of days, 9/30 of the month. So we ended up with an ending number that was actually below the FTE number. That's just because why the 2 numbers are calculated.

Okay. That makes sense. And then last one, as your overseas sales force goes back to the office by the end of the year, should we expect another step down in the sales headcount? Or is that a sufficiently small portion of the total sales force that it won't be as noticeable?

So first of all, it's a small portion of the sales force. Second, they are all NetCentric. Third, while there is some anti-vaccination sentiment in some of the countries, the acceptance of vaccines when available in Europe seems to be higher than the U.S. And we -- when we implemented voluntary return to office, we actually got a much higher take rate literally across each of our European markets. Now we have not yet tested the Singapore sales office as the government does not allow voluntary return yet. And then we had a bit of a setback yesterday in Netherlands where we went to voluntary, we actually had 100% participation in people returning to the office only to have an order by the Dutch government requiring companies not to allow people to work in the office until the end of the year, more than 50% of the time. So we're going to be in compliance with every local regulation, Europe is not monolithic. Spanish rules are different than Swedish rules. And we're going to comply. I actually had a call Friday with the U.K. teams and talked to them about just their sentiment. And there is a real increase in reported case volumes, again, in the U.K. And while there is not any governmental mandate against coming to the office, it may happen. So again, we literally monitor each of these jurisdictions daily.

Our next question comes from Walter Piecyk with LightShed.

Dave, that status is pretty incredible to 2/3 of your hires leaving. I'm just curious, I might be misremembering this, but I thought during the pandemic, you had talked about how people were as productive, yet some software working out of the office than in, maybe you said something differently, so you probably correct me on that, but why not just let them work from home and do their sales there? Why do you need to force them back in the office in order to maintain their jobs?

So what I had said previously of remote work is we quickly pivoted. We quickly modified our hiring and training, and we were as effective in hiring people remotely as we were in the office. Secondly, later in the pandemic, if you remember, it was August of 2020, we implemented a more -- a disciplined approach of managing out the underperformers, and we had several quarters of additional elevated sales force turnover and then it began to revert to normal. I think there are really 3 parts to answering your last question around remote work. Some employees left because they refused to be vaccinated. And the OSHA requirement doesn't give you a dispensation for people who work at home. If your company has 100 employees, you're supposed to require people to be vaccinated absent religious or medical exemptions. Two, there are a number of employees who have, I think, rethought their lifestyle and prefer to work remotely. That does not fit the Cogent model well. Again, to remind investors, our U.S. sales force is compensated hourly. We concluded that they are not exempt employees under the Federal Fair Labor Standards Act, and it's highly unusual to have remote workers hourly. And then the third point is that we have seen a difference in rep productivity in the office versus remote due to the ability to mentor those less mature reps by mature reps and managers physically in the same environment. While our reps could do their job remotely, their productivity was lower during the pandemic. Now some of that was as a result of the market, but some of that is a result of those reps not accelerating in their training as quickly as they would in the office. So most of our elevated turnover has been of the very newest reps. In fact, most of the turnover were reps that never had set foot in a Cogent office. And we believe that they will be more effective. We do have a program for tenured reps that do allow them to work remotely if they meet their sales objectives.

Got it. That kind of dovetails into my next question, which is, if that's somewhat reflective of the fact that this is not relatable to 2008, meaning that it's changed how people work, when your own employees, 2/3 of them are leaving, whether they're anti-vaccinated or it sounds like more of a lifestyle choice, then just in this -- I know that this is difficult to predict, you and I have been engaging on this -- on past calls. But when you talk about a couple of quarters away to get back to a consistent positive growth, do you think this business ever gets back to 2% sequential growth because that -- because your comment from the prepared comments was -- or maybe it was in the Q&A, a couple of quarters to get back to consistent positive growth, but does it get to 2% sequential growth, Dave, ever?

I think the answer is yes, Walt. And the reason is we have 16 years of history, and we know that we still have a significant amount of addressable market in our footprint that we are not serving. The value proposition that we deliver is increasingly differentiated from our competitors due to the fact that we're offering a symmetric service that is not oversubscribed and non-bought. If companies are more dependent on the Internet, they are more likely to buy from Cogent than a cable company or a phone company or a competitive provider due to the fact our services install faster and are 3x more reliable with 60x the throughput. So the value proposition is there. The issue that we are facing is twofold. One, there are less businesses in the buildings now, vacancy is almost 3x higher than it was pre-pandemic. And secondly, companies are struggling with the exact time line of their new normal. But as long as the buildings return to at least equal or better occupancy, our growth rate can easily achieve the 2% sequential that we've historically delivered.

Got it. And Dave, one last question. In the past, when we talked about gig service, I remember years ago, we used to joke about the fact that there was maybe some hedge funds that would buy gig service, and they would only use 12 meg, right? They were just over purchasing, but you were benefiting. But clearly, people understand the importance of connectivity now. And I would totally understand how all new customers are coming on a gig, but has your revenue been helped by the fact that over the past 2 to 3 years, you've had people that were 100 meg service effectively upgrading to gig service. And does that tailwind kind of peter out in the years ahead, so that ARPU benefit that you may have been getting from someone shifting from 100 mg or 200 meg up to 1 gig is no longer there?

So the answer is yes. We did have a tailwind, and that tailwind will diminish as the installed base increases. But there are 2, I think, more important offsetting factors. Ironically, we're actually starting to see an uptick in corporate customers wanting 10 gigabit connections such as your former employer, who is a Cogent customer, taking all of their 1-gig connections and upgrading them to 10 gig, again, recognizing the importance of bandwidth in trading operations. And then secondly, we have faced a 2-year headwind to VPN growth. Roughly, 25% of corporate revenues, 17% of total revenues come from VPN sales. Those sales basically stopped and churn increased as companies pruned their office footprint, but existing MPLS networks were not migrating to VPNs. We are now seeing a reacceleration of that VPN growth, which becomes, I think, even a bigger tailwind in terms of aggregate revenue than the 100 meg to gig upgrade cycle. So we end up, I think, net positive as people return to office. Sean?

Can I just add a little bit to that, Walt? I mean, from an atomic basis of how you're looking at how we're selling for a small premium, we're going to sell contracted service that is bidirectional 1 gig service to our customers. These are customers that have, on average, 8,000 square feet. They're paying close to $50 a square foot. So they might be paying $400,000 a year in rent. The idea that they're going to have service from a cable company or others that is not contractually committed to provide that 100 meg service. We are beginning to see that 1 gig is a big differentiator for a very small price. We can really dominate that service. And that is where the business is going.

Our next question comes from Michael Rollins with Citi.

So 2 questions to follow up. The first is, if you look at the NetCentric growth constant currency year-over-year, is there a way to unpack what's happening domestically, what's happening internationally as you've been expanding your reach in your network, maybe the impact, whether it's on revenue or profitability from doing more on-net? And I realize maybe it's hard to do because some of these things might overlap with one another. But just trying to appreciate how to unpack the growth and think about maybe some of those factors in the future. And then just secondly, you mentioned that the gross adds are down in corporate relative to pre-pandemic levels. Just curious how much of it do you think is the pandemic. And is there a risk that some of that erosion of gross adds could be competitive, just as other companies could be using this opportunity to invest or try to get more aggressive at the market that you focus on?

Yes, sure. So let me take the NetCentric one first. We have benefited from multiple underlying trends. Growth in NetCentric revenue is driven by increase in traffic, increase or decrease in price per megabit and the percentage of traffic that is two sided. We have seen outsized NetCentric growth after a period of underperformance in NetCentric due to the factors of smaller customers growing faster than the bigger customers, giving us a more diversified customer base; two, the percentage of traffic that is both originating and terminating on the Cogent network. So it's why our revenue growth at 25% was equal to our traffic growth, whereas in a completely normalized world where these other trends are not taking place, and there's a 23% year-over-year price reduction, we would actually see flat revenue. So we have benefited. In terms of internationalization, there's a couple of parts to answer that. [ Content ] is still predominantly a U.S. product that's exported to the rest of the world. That's not totally true, but it's more heavily U.S. centric. And then on the access side, the U.S. has more access networks that are global peers and less paying customers as a percentage of the access space. We have many access customers in the U.S., but globally, most of the international markets buy Trans. So we get a higher percentage of 2-sided traffic. Now the traffic may actually -- the [ content ] may originate in the U.S. and terminate in Asia or it may actually be European hosted, but then terminating in South America. It's totally dependent on language. Again, we think of the Internet is an English language phenomena. But you have to remember, the vast majority of people that use the Internet don't speak English and yet Internet traffic growth is growing.…

Yes. Let me just add a little bit to what Dave is saying. We were nervous about what we have had elevated churn during the pandemic. So we went and looked at a series of buildings, which had higher levels of churn, the Graymark building, for example, in New York, which we saw 8 circuits leave. They were, for the most part, all FE that is 100 megabits per second. And we asked them 4 questions as we do: Did you move? Did you close the office? Did you got a business? Or did you got a competitor? In the Graymark building, all of them were one of those answers except for going to a competitor. And I think anecdotally, we do see on the 100 meg side, I think it was a retailer that did doughnuts that was competing with us, we're competing with a cable operator and the price got so low. I think Dave didn't approve the pricing. That's -- in the low end, we do see that where it's very price competitive. On the 1 gig side, we think we're in very, very fine shape.

Our next question comes from Frank Louthan with Raymond James.

It's Rob on for Frank here. So just a quick one from me. So when do you guys see the bottom in the off-net disconnects? And are you seeing any new pockets of off-net demand that could potentially offset that as population and work shifts have sort of rebounded?

Yes. So a positive for the off-net business is the fact that over 55% of the square footage in North America office space is now available with fiber provided services from one of our 90 different vendors. So we have over 4 million square feet that we can sell into. That's the positive. The negative is that it's very expensive to sell just off-net. The competitive differentiation is not large enough. So our off-net sales are almost always catalyzed by an on-net relationship. We also know that because there are oftentimes multiple off-net choices per endpoint in those 7,500 buildings where we sell those 12,500 connections, we have the ability to compete them against one another and drive down prices. So ARPUs have gone down. I think as we look at off-net, it's almost exclusively corporate. As the corporate on-net rebounds, the corporate off-net will rebound and companies will take secondary connections and therefore we'll get back to growth that more or less mirrors unit-wise, the on-net growth and the only drag will be ARPU. So rather than kind of being a 2% grower, it's kind of like a 1% to 1.5% sequential grower.

Yes. I would just add into that. And again, this is the Gigi theme, but we rolled out Compass, which is our CRM system last year. We created an ability for salespeople, particularly new salespeople to get between 30 and 50 leads every morning. This would be a potential customer in a building that we service through the 4 million buildings that Dave spoke about. And we are finding real success. First of all, it's really good training for the new sales people. But that product, the Gigi off-net, despite its pricing, which Dave talked about, we are up over 20% year-on-year in that unit growth. So that product, inside that, we're having challenges on VPN, as Dave talked about. But for that product itself, we are seeing very, very strong growth.

Our next question comes from Brandon Nispel with KeyBanc Capital Markets.

It's Evan on for Brandon Nispel. Two questions. Are you guys seeing any impact to your customers related to supply chain or inflation that could potentially flow into impacting you guys? And with the VPN, are you seeing any decision changes from your customers for taking an alternative product? Or are they still pretty set on taking VPN?

Yes. So 2 very different questions. Fortunately, we do not have much exposure to manufacturing as a sector. Virtually all of our customers are white collar in an office building by definition. And there, I think there's been much less supply chain issues. And as Sean pointed out, with the average customer spending $400,000 on rent for a location, spending $6,000 a year on Internet is a trivial expense. And while they may experience inflation in other areas, technology has always been deflationary, and I think customers expect that. And we deliver on that by delivering bigger pipes at a much lower cost per megabit, whether it be the migration from 100 to gig, 1 gig to 10 gig. So I just don't see a lot of supply chain issues really impacting us. In terms of the VPN question, there, I think, companies have kicked the can down the road on MPLS replacement. They know it's antiquated, but they are waiting for 2 things: one, better VPN solutions, which may or may not come from the vendors. But two, exactly what locations they want that VPN to work in as they figure out how many branch offices they will close. I do think we're starting to see a reacceleration and interest in VPNs, and most of those are either SD-WAN or VPLS, and there's really not a lot of alternatives. And as I stated earlier, we've seen the additional benefit of companies now adding an extra node that didn't exist previously for their remote work VPN concentration in data centers. So all in all, we think that we'll see VPN business reaccelerate the growth rates similar to DIA.

I'm showing no further questions at this time. I would now like to turn the conference back to Mr. Dave Schaeffer.

I'd like to thank everyone for their interest. I know it's a long call, but a lot of excellent questions. Take care, everyone. Stay safe. Thanks. Bye-bye.

This concludes today's conference call. Thank you for participating. You may now disconnect.