Good morning, ladies and gentlemen, and welcome to the Cogent Communications Holdings Fourth Quarter and Full-Year 2019 Earnings Conference Call. At this time, all participants are in a listen-only mode. Later we will conduct a question-and-answer session and instructions will follow at that time. [Operator Instructions]. As a reminder, this conference is being recorded and will be available for replay at www.cogentco.com.I would now like to turn the conference over to your host, Mr. Dave Schaeffer, Chairman and Chief Executive Officer of Cogent Communications Holdings.

Thank you and good morning everyone.Welcome to our fourth quarter 2019 and full-year 2019 earnings conference call. I am Dave Schaeffer, Cogent's CEO, and with me on this morning's call is Tad Weed, our Chief Financial Officer.We're pleased with the results for the quarter and for the year and continue to be optimistic about the underlying strength of our business and the outlook for 2020 and beyond.Our quarterly gross margin reached a company record of 60.3% increasing sequentially by 40 basis points and increased by 230 basis points from the fourth quarter of last year. Our gross margin for full-year 2019 increased by 190 basis points from full-year 2018 to 59.9%. Our EBITDA margin for the quarter increased by 160 basis points from the fourth quarter of 2018 to 37.6% and increased by 70 basis points to 36.2% for the full-year 2019 versus 2018. On a constant currency basis, we achieved sequential quarterly revenue growth of 2.5% and year-over-year quarterly revenue growth of 6.8%.Our year-over-year quarterly traffic growth was 30% and we achieved sequential traffic growth of 9%. For full-year 2019, traffic on our network grew by 35%.During the quarter, we returned $29.8 million to our shareholders through our regular quarterly dividend. As posted on our website, 54% of our $112.6 million total 2019 dividends should be treated as a return of capital, and 46% should be treated as taxable dividends for U.S. tax income purposes.We did not purchase any common stock in the quarter. At quarter's end, we have a total of 34.9 million available in our stock buyback authorization and that authorization will continue through December 2020.Our gross leverage ratio in the quarter decreased to 4.86 from 4.97 the previous quarter and our net leverage also decreased to 2.86 from 2.92. Cash held at Cogent Holdings at quarter-end was $110,300,000. This cash is unrestricted and available to be used for dividends, buybacks, or both. Cash held at our operating company at quarter-end was $289.1 million, giving Cogent a consolidated cash balance of $399,400,000 at year-end.We continue to remain confident in our growth potential and cash generating capabilities of our business. As a result, as indicated in our press release, we announced yet another $0.02 sequential increase in our regular quarterly dividend from $0.64 a share per quarter to $0.66 per share per quarter. This represents our 30th consecutive sequential increase to our regular quarterly dividend. Throughout today's discussion, we will highlight several operational statistics. I will review in greater detail certain operational trends and Tad will provide some additional details on our financial performance. Following our prepared remarks, we'll open the floor for questions and answers.Now I'd like to turn it over to Tad to read our Safe Harbor language.

Thank you, Dave, and good morning to everyone.This earnings conference call includes forward-looking statements. These forward-looking statements are based on our current intent beliefs and our 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 comparable and GAAP measurement in our earnings release which is posted on our website at cogentco.com.Now I'd like to turn the call back over to Dave.

Hey, thanks, Tad.Hopefully you've had a chance to review our earnings press release. Our press release includes a number of historical metrics. Now for some comments about our performance against expectations and targets. Our corporate business, which represents 69% of our total revenues for the fourth quarter, our corporate business has been growing above our targeted guidance, long-term full-year range of 10% and in fact grew 10.1% from the fourth quarter of 2018.Our NetCentric business, which represents 31% of our revenues, has been underperforming compared to our historical averages and declined by 1.4% from the fourth quarter of 2018.Our corporate business grew sequentially by 2.5% as compared to the previous quarter. The impact of foreign exchange primarily impacts our NetCentric business. On a constant currency basis, our NetCentric business increased on a year-over-year basis for the fourth quarter by one-tenth of a percent and increased by 2.6% from the third quarter of 2019.Our quarterly cash flow as defined by EBITDA minus CapEx minus principal payments on our capital leases. For full-year 2019, our cash flow grew by 14.1% from 2018. Due to the excellent operating leverage of our business, we expect our cash flow to continue to grow at similar rates. Our long-term EBITDA margin expansion targets guide to an annual improvement of approximately 200 basis points per year. On a multi-year constant currency, long-term revenue growth target remains approximately 10%. Our revenue and EBITDA guidance targets are intended to be multi-year targets and are not intended to be specific quarterly guidance or even specific annual guidance.Now I'd like to turn it back over to Tad to give you some additional details on some of our operating results.

Thanks, Dave, and again, good morning to everyone.And I'd also like to thank and congratulate the Cogent team for their results and their continued hard work and their efforts during another very productive quarter for the company, and also a very productive year for Cogent.Some comments on revenue by corporate and NetCentric and customer connections. As a reminder, we analyze our revenues based upon network type, which includes on-net, off-net, and non-core, and we also analyze our revenues based upon customer type. We classify all of our customers into two types: NetCentric customers and corporate customers. Our NetCentric customers buy large amounts of bandwidth from us in carrier-neutral data centers. Our corporate customers buy bandwidth from us in the large multi-tenant office buildings typically in North America.Some comments on revenue and customer connections by customer type. Revenue from our corporate customers for the quarter grew sequentially by 2.5% to $96.8 million and grew year-over-year, as Dave mentioned earlier, by 10.1%. Revenue from our corporate customers was $373.7 million for the full-year, which represented an increase of 10.6% over last year. At the end of the year, we had 48,486 corporate customer connections on our network and that was an annual increase of 7.3% over the fourth quarter of last year.Quarterly revenue from our NetCentric customers increased sequentially by 2.4% to $43.5 million, but declined year-over-year by 1.4%. Revenue from our NetCentric customers was $172.5 million for full-year 2019 which was a decrease of 5.4% over last year.On a constant currency basis, our quarterly revenue from our NetCentric customers increased however by 2.6% sequentially and increased year-over-year slightly by 0.1%. We had 38,053 NetCentric customer connections on our network at quarter-end, which was an increase of 9% over the fourth quarter of last year.Our NetCentric revenue growth experiences significantly more volatility…

Hey, thanks, Tad.Now for a couple of comments on the scale and scope of our network. We have over 957 million square feet of multi-tenant office buildings in North America directly connected to the Cogent Network. Our network consists of over 35,520 metro fiber miles and over 57,600 intercity route miles. The Cogent Network remains the most interconnected network in the world with direct connectivity to 6,950 networks. Less than 30 of these networks that connect to Cogent are settlement free peers, with the remaining over 6,920 networks, being paying Cogent transit customers. We're currently utilizing approximately 29% of the width capacity in our network. We routinely augment this capacity as portions of our network need those augmentations to maintain these low utilization rates.For the quarter, we achieved sequential quarterly traffic growth of 9% and year-over-year traffic growth for the quarter of 30%. We operate 54 Cogent controlled data centers with a total of 617,000 square feet of rays for space, which is operating at approximately 31% capacity. In 2019, we did add two additional data centers to our footprint. These are Cogent controlled facilities increasing from 52 to 54 at year-end.Our sales force turnover at 4.9% in the quarter was again better than our long-term rep turnover rate of 5.6%. This is an area that we will continue to focus on with training and rep improvement. Our quarterly sales rep productivity was 4.1% or 4.1 units per full-time equivalent rep per month. The productivity rate is below our long-term historical average of 5.1 units per FTE per month. Our sales rep productivity was impacted by the decline in our average rep tenure and significantly impacted by the fact that we've been selling a significantly larger number of large ports, allowing customers to aggregate traffic and not have as many…

[Operator Instructions].Your first question comes from Sami Badri from Credit Suisse.

Hi, thank you. Dave maybe could you just -- well more perhaps maybe take a step back at a high-level question is you've always referred to the big or the next big app that could really come onto your network. And I think we've been hearing about a couple of new apps that have been coming into the broader global network, would you say that next big app that could take Cogent to the next level is now here and beginning its commencement through your network? Or would you say there's still something else that could come through that as much bigger?

So I think the Internet while approximately 30 years old, and has grown at a compounded rate of 25% has been able to avoid the law of large numbers and slowdown due to new applications. Over the past year, we've seen a proliferation of over-the-top direct-to-consumer products that is dramatically increasing viewership on the Internet. We think over the next several years, the majority of minutes of video will be consumed via streaming services. Those streaming services are also becoming higher resolution. So we feel very comfortable that this single application will continue to drive compounded growth for the Internet and therefore Cogent as the low cost provider will disproportionally capture that growth and grow even faster. And in fact, our growth rate year-over-year at about 35% indicates that.Finally, I think there are many new applications that are being developed, different forms of video, more interactive video, video that will have augmented or virtual reality embedded in it. And it is these higher resolutions that I think will provide yet another next leg in Internet traffic growth. So we remain extremely positive on the long-term prognosis for accelerating traffic growth. And as we've seen competition deemphasize transit as a product, we're increasingly gaining market share. And I think all of this is now reflected in the fact that on a sequential basis, our NetCentric revenues improve materially, almost to our long-term growth targets. And then on a constant currency year-over-year basis, we return to positive growth after a period of negative growth. So we feel very confident that these trends will continue over the next several years.

Got it. Thank you for that color. And then so my next question has to do with sales force productivity. And maybe I was hoping you could expand on some of the comments you just made regarding the dynamics that were there. It's just that this sales force productivity number that you just reported was probably one of the lowest in the company's history. And at the same time, your revenue growth rate year-on-year actually accelerated, I know you mentioned a couple of factors. But can you just kind of expand on what's going on here? Just because this is a new dynamic that's being introduced into the model and the dynamics that we may not necessarily be very well versed into?

Yes, fair, Sami. So there are two different things going on. The first is across the entire sales force, where we accelerated hiring, and therefore the average tenure of a rep declined. And we think will revert back to kind of a normal rate of hiring in 2020 our goal was to hire between 7% and 10%. Last year it was a 13% increase, which was outsized and therefore tenure came down.But the more significant factor in driving the divergence between unit productivity and revenue growth has actually been the sale of larger ports. The reason we give the metric of unit productivity is it's the only common measure across the three different groups within our sales force. That being our NetCentric reps, our corporate reps that sell to larger multi-site customers, and our reps that sell to smaller single site customers. So we try to come up with some kind of uniform metric. But what has happened and for the second quarter in a row, the majority, more than 50% of our corporate sales had been GigE interfaces. And that is why our corporate ARPUs actually increased sequentially in the quarter.Secondly, on the NetCentric side, we're seeing a rapid adoption of 100-gig interfaces. This is actually the third time in Cogent's history that we've seen grooming going on with NetCentric customers. So 15 years ago, most NetCentric customers took Gig interfaces. There was a migration in the kind of 2006 through 2008 period to 10-gig interfaces where customers consolidated multiple, one-Gig interfaces into fewer 10-gig interfaces.Over the past year, as the equipment vendors have brought down the port prices on 100-gig interfaces, we were seeing most NetCentric customers choose to use 100-gig interfaces. So on a unit basis, they may be selling fewer connections, but are in fact able to generate more revenue with higher committed bandwidth rates and higher usage base bandwidth. So one 100-gig connection may be replacing four, five, or six 10-gig connections. This is accelerated by the fact that most carrier-neutral data centers charge a recurring fee for cross connects and customers are constantly looking to reduce their cost basis in delivering services. And if they can groom cross connect they can in fact lower their effective cost per megabit.So I think we've still got ways to go in this trend. I know equipment vendors are today testing 400-gig interfaces. I think we're probably still several years away from that being commercially available on a cost effective basis. But I do think for the next several quarters, we will continue to see a proliferation of 100-gig interfaces, with NetCentric customers, and 10 -- and 1-gig interfaces with our corporate customers.

Got it. Thank you for that clarification. And then my last question has to do with the fact that your adjusted EBITDA margins are clearly delivering expansions, your cash flow is clearly growing. And then the one thing is that your dividend growth rate necessarily isn't necessarily growing, it's actually decelerating now. Is there a scope at some point for you to consider increasing the step-ups in the quarterly dividend from $0.02 to something higher and can you just give us maybe an idea on how you're thinking about capital allocation from there?

So, that capital allocation discussion is probably the most important discussion at each of our quarterly board meetings. We had a meeting yesterday where that was a major topic of discussion. Just arithmetically as the base gets larger, and the unit size of the increase is $0.02. To refresh people's memory, at one point, we were increasing at penny a share and then we increase -- the rate of increase to $0.02 a share. That will arithmetically decelerate. It is something the board is very cognizant of. They took absolute note of the fact that our gross leverage and net leverage ratios declined in the quarter and are below the mid-point of our net leverage target. So it is something that is under active consideration. But I think for this quarter the board felt staying the course due to a lot of the macro uncertainty made the most sense.

Your next question comes from Philip Cusick from JPMorgan.

Hi, good, thanks. Thinking about NetCentric again, it seems like this is stabilizing at least, are we may be bottomed on the impact of customer concentration?

I think so, Phil, because we're seeing a number of new OTT players compete for consumers eyeball time, and therefore the base is starting to broaden. I'm not trying to imply that any one service provider is necessarily suffering. But I think the growth in the market is now being spread out among a half a dozen different players as opposed to one or two major players. And that means there's less monopsony power on the part of that customer. And therefore, we see the effective rate of decline on a volume weighted basis moderating. Even though the nominal rate is continuing to decline pretty much in line with historical averages, we're seeing a lot of growth at a faster rate than the average from some of these new entrants.

Do you think that on a sequential basis, currency adjusted this could continue to grow in the next 12 months?

It appears to be the case to remind investors this is a usage-based business. Even our customers don't have exact visibility into how many minutes a day each of their end users is going to be spending.But when we look at our NetCentric sales funnels, the roughly 180 NetCentric reps, they appear to drive port sales that will ensure a growth in revenue. We see on a customer-by-customer basis very encouraging signs of traffic growth. But again, we can't predict, the consumer reaction to different products or series that are outlined by the various different providers. We think they'll all be successful. I know there's been some providers, household names that have had more a linear model that now we're going direct-to-consumer and have had just tremendous uptick. And we think that's going to continue. So a long answer to your question, we think we're going to see sequential improvement, but it is just not something that's totally in our control.

Okay. And then following-up again on the reps, given the growth in that sales force, should we expect as large or maybe even larger sequential pressure on margins in the first quarter?

So you will expect EBITDA margin pressure in the first quarter due to increased SG&A on a seasonal basis, but not on a year-over-year basis when looking at the quarter. Gross margins will continue to improve at kind of the similar pacing we have had. SG&A expenses pickup in the first quarter for the reasons that Tad enumerated. We spent about $1.2 million on our sales kickoff meeting. That may sound like a small number, but to a company like Cogent that's a meaningful, extraordinary expense. The bulk of our audit expense falls in this quarter, and we have the reset on social benefits and our payroll resets, employee, employer taxes. All of those have a bit of a drag on margin. So you should expect a step-down in the EBITDA margins, Q4 to Q1, as we have every year, but when compared Q1 of 2020 versus Q1 of 2019 we will be doing better.

But no bigger than typical seasonal step down?

That is correct, Phil.

Your next question comes from Colby Synesael from Cowen.

Great, thank you. Just a few first-off on the dividend, you have a longer-term goal for revenue plus 10%, EBITDA margin of plus 200 basis points, curious if it makes sense to have a longer-term dividend objective or goal maybe plus 10% as an example. And have you or the board considered that? Second, did I hear correctly perhaps in response to Sami's question that you anticipate traffic growth being above the 30%, we just saw here in the fourth quarter when we look at back on 2020, and then it gets driven by the OTT guide? And then the last question tied to that. Can you remind us how the contract structure works for these OTT deals that you're signing, are these month-to-month or these year-to-year? Should we be thinking that there could be maybe a year from now, some significant step down just trying to get a sense how to think about the pricing negotiations? Thank you.

Sure, Colby thanks for all the questions. Let me start with the return of capital question and the dividend question. We are very cognizant of our leverage. We realized that our underlevered balance sheet is an asset that we need to put to work for our equity holders, and we're constantly evaluating the debt markets, which would give us additional flexibility in accelerating return of capital.Secondly, as we have publicly commented in the past, we are somewhat agnostic to dividend versus buyback as the mechanism of return, looking at the relative efficiency at a given point in time. We fully acknowledge that without adjusting either the rate of buybacks or the rate of increase in the dividend, we will fall below our leverage targets and not be returning the optimal amount of capital.I think to kind of target a specific rate of growth in dividend is probably not optimal. I think we want to maintain the flexibility of using both tools in our toolbox. But as I said in the response to Sami's question, there was a vigorous debate at the board level and that will continue and be the primary focus of the board for each and every of our board meetings, and we acknowledge that we will either have to step up the pace of buybacks or step up the pace of dividend growth, or arithmetically will fall below our targets. Again, these are good problems to have compared to many companies that were often compared with.Now with regard to traffic growth, we believe that there is a long-term secular trend of OTT that is accelerating. Cogent has relationships with almost -- well every major OTT provider and the vast majority of access networks globally that are the recipients of this over-the-top traffic. We expect that to continue…

Your next question comes from Walter Piecyk from LightShed.

Hey Dave, how is it going?

Good, Walt.

When I talk about NetCentric, I think on our mass, it looks like this is the first time you've returned to annual growth for like nine quarters. Is that sound, right, year-over-year growth?

That is correct, yes that is correct, Walt.

And then if you look back like, I don't know, whatever X number of years, however model goes back there seems to be like an ebb and flow in the sequential, the sequentials are obviously a little bit more interesting to us. So I'm just curious if you can give us kind of little bit more sense of what drove I think, yes, like 2.5% sequential NetCentric this quarter, whether it's geographic color, or type of customer color, anything. And then also, as part of those comments kind of wrap in thoughts on sustainability. And again, we had what four, five quarters of negative sequential, but again, if you look back over six or seven years, it's like you're up a couple of quarters, you're down a couple of quarters, just thought process around what drove this one and what you're thinking going forward?

Yes, so I think the key driver on NetCentric remains OTT video. There were a couple of very high profile launches in the past six months that have driven material acceleration in revenues for NetCentric. And we win two ways. We win by carrying traffic for those OTT players. But we also win because of the 6,920 access networks around the world that whose customers download more content from those players. So I think the broadening of the universe of OTT players has done two things, it's accelerated OTT in general, and it's allowed us to not be as dependent on a few names and therefore at the very lowest price point.We've seen multiple instances in the past where customers hit a inflection point in their business, we benefit for several quarters and then their rate of growth starts to mature. So the NetCentric revenues, which are highly usage dependent, are very hard to predict over the short-term. Over the long-term, the cycles that you refer to are actually pretty easy to predict. And I think it's important for investors to step back and look at the bigger trends.Also on the corporate side, which tends to be this very granular business of small connections there we see a much more linear progression of growth.

Yes. And corporate is obviously very predictable. But on this one, I mean I guess you could argue that word inflection point, cord cutters are kind of -- this is the point, right? So, is it possible that you can just sustain sequential growth in NetCentric going forward? Do you think we're still going to have some of these, kind of three quarters, all three quarters on type of situation?

No, it is really too hard to predict, Walt. I think, we -- after you pointed out have had nine poor quarters, and now a positive quarter, I think we're in for a period of better performance. Will it be permanent? We just don't have -- early into that, what we can be very confident though is that if the market grows, we will grow faster than the market. We will capture a disproportionate share. And that disproportionate growth comes from the fact that we price at a lower price point.

Got it. One last question, Dave. I mean, people are looking for shelter on this, the selloff based on coronavirus, is there any -- have you seen any increase usage in areas that have been affected in terms of traffic? I mean, is this something that that could stimulate traffic if people are not going outside and using the Internet more in any of these given market any, any learnings there in the past couple of weeks?

So obviously, logic says that if people are staying home, they're looking for things to do and OTT video is a low cost and virtually unlimited way to fill up to that time. Obviously, China's been the hardest hit. We are the main provider to the three major state sanction networks in China. But they, the Chinese government kind of throttles what goes in and out of the country. So it has been a little hard for us to determine exactly how much of the growth and traffic from those three providers is directly related to coronavirus -- coronavirus. I do think as the fear of contagion spreads, we will see some traffic benefit from that, we wish.

But you haven't seen anything in Spain as an example, in the last couple -- but Dave you haven't seen anything in Spain as another example in the past couple of days?

We've seen slight pickup in traffic over the past weekend. But again, I just I've seen pickups in the past and I'm not going to take three days and attribute it to a single cause, I just don't have enough predictive data.

Your next question comes from Frank Louthan from Raymond James.

Great, thank you, got one strategic question and I would more of a practical one. On the strategic side, from the peak, the NetCentric business is down about 40% as a percentage of your revenue from where it used to be. And kind of given that that pretty consistent trend, and then how well you've done on the corporate side, should you adjust your strategy to become much more of a corporate focused business going forward, to drive growth and just let the NetCentric kind of do what it's going to do and perhaps with self-provisioning and other things impacting you, how do you think about that from a strategic standpoint? And then secondly, I apologize if it's addressed in the remarks. But the asset sales picked up a bit. Are we back to dusting-off some of the equipment from the supply room and selling that and how much -- how should we think about that as how it's contributing revenue and cash flow for the remainder of 2020?

Okay, so two very different questions. Let's start with corporate versus NetCentric. There is kind of a natural skewing, in other words, we go where the demand is and corporate demand is driven heavily by the adoption of cloud, SAS, and other OTT applications, the biggest being SD-WAN and VPNs. We are growing both parts of our sales organization.NetCentric has always ebbed and flowed as part of our business. We remain committed to that market and continue to focus sales resources. In fact as it was pointed out earlier, it is growing on a sequential basis. And more importantly, we are gaining market share. Both customers create a network effect for Cogent where our corporate customers benefit from the number of NetCentric customers and NetCentric from corporate. More of the growth in our headcount has been on the corporate side because there's quite honestly just more addressable market and more marketing for us to do. The NetCentric market is one, where Cogent is fairly well known. We are the second largest provider in the world, and virtually all prospects know of us, now often times they still need sales support to help them do more with us.We think both of these businesses can be meaningful contributors to free cash flow going forward. Our process always begins with an outbound telesales effort for both corporate and NetCentric. This is not a self-provision product. So I don't see that materially impacting our growth in either business.Finally, to your question of asset sales, I will let Tad touch on that.

Yes, sure. So if we look back, I'm looking back through the first quarter of 2018, we have not had an asset sale amount more than $500,000. So the amounts really have been de-minimis since really since the beginning of 2018, the largest amount in that one quarter was $500,000 in the first quarter of 2019. So they have not been material as of late but it could change in the future but currently that they've -- they've really not been material.

Right, I mean it's really dependent on the markets need for the equipment that we view is obsolete to Cogent. We have huge inventories of equipment and sometimes they're buyers, sometimes they're not. And finally, that revenue or that cash that we received is not booked as revenue because it is not Cogent's core business, we report service revenue but the gain is reflected in EBITDA but and it is cash but we don't actually book that as revenue.

Your next question comes from Nick Del Deo from MoffettNathanson.

Hey, thanks for the question. I'll just ask one given the time, if I look at your total network operations expense ex-USF, it was down a little over a percent full-year 2019 versus full-year 2018 despite the expansion in your network and because I'm talking dollars, not percent of revenue, how much of that decline was a function of FX, what are the other drivers we should be aware of and what should we expect going forward for that line item?

So roughly 25% to 30% of the network resides outside of the U.S. and therefore FX would impact that third of the expense to the extent that the dollar has strengthened, therefore on a reported basis would be a lower expense.The second big network expense area is actually the cost of off-net circuit. And we have been able to negotiate with all of the major loop providers in North America, substantial discounts and large part because of the aggressive competition between Telco and cable for providing wholesale point-to-point services for buildings that we have not brought on-net. So we have about 1.6 million buildings in North America. We have 90 different vendors. But the reality is buy our material and we've seen an average decline in the price per loop for a comparable loop of about 8%. So a big part of why the ARPU decline on off-net has occurred is because those loop prices have come down and we're roughly at a 50% margin in that service.Now that's partially offset by the same trend that we're seeing in on-net corporate, which is a move towards larger connections. So for a long-time, we were supporting sub 100-Meg Ethernet based off-net services. Today, we only do those on a special assemblage. Most of our sales are still 100-Meg but we expect the trend like we're seeing in on-net to bleed over to off-net and eventually gigs will become more standard.Finally, we've seen a number of new subsea cables to be constructed. And those cables put pressure on that market. And we concluded long-ago that renting versus owning there made more sense. In fact, Cogent did own some cables earlier in its history through various acquisitions that we abandon and today remain a effective leaser of capacity.

Your next question comes from James Breen from William Blair.

Thanks for taking the question. Can you just talk about the impact you're seeing in the SD-WAN market and the VPN market, I know you guys moved in there little bit more and how that is sort of progressing going forward. Thanks.

Yes, sure, Jim. Thanks for questions. So just to remind investors roughly 25% of our corporate ports sold are for VPN purposes. About 75% of corporate ports are for dedicated internet access. We see a large number of MPLS customers begin the sales conversation around SD-WAN. Most SD-WAN sales have actually turned into VPLS sales; we have and continue to sell SD-WAN it becomes a critical part of the sales discussion. But when customers understand the limitations of the current state of customer premise equipment, they typically choose to go for VPLS to get higher throughput rates.Now we're seeing improvements in equipment. I think the reality of that CPE equipment will catch-up to the hype over the next year or two. And we'll see a broader deployment of true SD-WAN. But I think the key takeaway is, customers are increasingly fed up with MPLS and are willing to switch and they'll switch to either product. Today, most of it is VPLS. And we think that the MPLS market in aggregate has already declined about 20% from its peak.

Your next question comes from Michael Rollins from Citi.

Hi, good morning. Just two follow-ups if I could. The first is just in terms of the USF changes that have helped a little bit on the revenue line? Is there an expectation that those rate could actually decrease in the future? Is that something that we should just be keeping in mind as the government is always rebalancing those rates periodically? And then, secondly in terms of some of the pricing metrics that you disclosed on the average price per new contract going to 28 Meg, how does that impact kind of the flow of revenues as we look at over the next 12 to 18 months? Thanks.

Okay. I'm going to let Tad take the USF and I'll take the pricing push.

Yes. I guess similar to foreign exchange; the rate of USF, the tax rate is entirely outside of our control. And it is directed each quarter by the government. Well, recently it has increased, however it is going to decrease and it is scheduled to decrease in the first quarter. So it's kind of out of our control recently for example, for the fourth quarter it's about $4 million. There was about $4 million, that's the same amount for the third quarter. There was about $3.5 million for the second quarter, because the rate had increased. But the rate has been scheduled to decrease for the first quarter of 2020. It's established by the governments and it's literally outside of our control.

Yes and maybe just a little background of how that works. So our VPN services are subject to USF. The government sets a target amount of money it wants to distribute through Cat 2 or Big Tap or E-Rat programs. And then reverse engineers the applicable base of services to target how much they wanted to raise and then project a new rate that is adjusted quarterly. We pass that through on a gross basis, it actually negatively impacts margin, but it does obviously help revenue because revenues we reported inclusive of that USF fee. I would suspect as the government really tries to accelerate broadband deployment in rural areas. In aggregate this rate is probably going to go up over time as the base is shrinking, and the amount of money needed is going up, but on a quarterly basis, it's really impossible for us to predict.Now to the flow through of discounting. Customers consistently look to optimize their spend; this is really a NetCentric phenomena. As I mentioned earlier, the typical customer signs a three-year contract. But because of outsized growth in their business during that term, they'll come back and renegotiate. And when they do, it's usually for a larger commitment it's at a later point in time and it will re-extend the contract for another longer period since they're partially into their period. This phenomena has been very consistent for the past 10 years or more. We've reported that number consistently and it hasn't materially changed. So I think the decline in the rate of new sales this quarter is pretty much in line with historical trends.The installed base actually declined a little less rapidly than historic average, the new sales declined a little faster. But if I go point you back to two or three quarters ago, those two parameters were almost completely inverted where new sales were declining less rapidly and the base was declining more rapidly. There's just some oscillation based on when customers renegotiate these contracts.

And much like you do for FX commentary about the expectation like sequentially or year-over-year in the coming quarter, is there some help in terms of the amount of change that that USF factor would impact revenue in the calendar first quarter?

Well, first of all, we only have visibility after USAC; the FCCs contract establishes the new rate, which changes every quarter. So annual projection would be meaningless on our part. And I think the materiality of this; just from where we sit to this quarter is not meaningful. I mean, it'll be down a little bit, but it's not a material number.

And just as a reminder, again it's a U.S. based tax, only U.S. based and it's only based upon part of our services that it's been applicable to so. But –

I mean you might be talking.

I don't know first of all. I don't know first of all --

A couple of hundred thousand dollars maybe, I mean it's just not a meaningful number.

What the FCC is going to do with their USF rate, it's just -- I guess I wasn't surprised when it increased, but I was frankly surprised when it recently decreased. So it's --

Thank you.

Yes, it's just not many -- not many.

Your next question comes from Tim Horan from Oppenheimer.

Thanks. Dave, you're adding a ton of sales people, obviously. And I'm assuming you think you can gain a lot more market share in both your market segments. Can you just talk about maybe where you think you are with market share on both sides and what that's kind of trending at? And I guess specifically on the NetCentric side, it seemed like the CDNs grew a lot faster than the score than you did. Do you think they're hurting you at all in terms of your growth and market share there in delivering traffic or just how you think about it? Thanks.

So, we're adding sales people to both corporate and NetCentric. I would say that the rate of increase is slightly greater on the corporate side than NetCentric. So on the corporate side, we start with the buildings we're in and the number of connections and number of unique customers we've sold per building. And we've sold 23 connections per building to about 15.5 customers per building. So out of 51 potential customers were roughly 30% penetrated. We're adding new buildings.In the off-net corporate market, we have less than 2% market share. But we also know that without an on-net relationship, the cost of sales are prohibitively high. So we have no real desire to go after just off-net only corporate customers. And increasingly, as the remaining customers in our footprint who have not bought from us deploy new technologies like cloud and SaaS, they become right to switch to Cogent and that's why our rate of penetration per building continues to improve almost linearly, even overbuilding so depend on-net for more than 10 years. So we feel that we have a long runway to grow.On the NetCentric side, our market share can be measured three ways. We have roughly 27.1 connections per data center. Secondly, we have about 22% of the bits generated. And we have about 13.5% of the dollar spent on transit services. We're gaining share under all three of those metrics, selling more customers per data center, getting a bigger share of traffic, and most importantly, a bigger share of revenue.To your point about CDNs, every major CDN is a Cogent customer. We sell bandwidth either directly to companies who build their own CDNs or to third-party CDNs that then incorporate our bandwidth into the service they are selling. So CDNs are not competitors but rather a portion of our ecosystem and a customer base for Cogent.

That's very helpful. Any update maybe just from your response, buildings over 10 years, can you maybe give some color on the penetration there, ultimately where you can kind of get to? And now, do you think the CDNs did grow fast in this quarter? I mean, it seems like talking to the three or four that we know, it seems like they did grow bit faster and clearly revenues accelerated a bit more for them.

Yes and I think some of that acceleration in Cogent's revenue growth came from those same CDNs. Some of it came directly from end-user customers. And CDNs incorporate space, power, processing, storage, and bandwidth as a bundled product. We think we will continue to win a disproportionate share of revenue from them.In terms of penetration, we think on the NetCentric side that we can eventually get to about a third of the market, particularly as a number of other historical players continue to deemphasize transit, due to the rate of price decline and the commoditization.And I think as Cogent has proven over 15 years as a public company, two things. One, the depth of the Internet has been greatly exaggerated and is not -- that's going away. And then secondly, even in a world where prices declined 23% per annum compounded, we have demonstrated our ability to continue to grow our revenues and most importantly and grow our cash flow against that. So we feel very comfortable that we're committed to the transmit market and our penetration is only going to increase.

Your next question comes from Bora Lee from RBC.

So following up on your comment about more than 50% of corporate sales being Gig interfaces, what's the mix of new sales versus swapping out of lower bandwidth to existing connections? And then secondly, can you give a sense of the pricing differential between Gig ports and the most commonly taken lower bandwidth corporate connections?

Yes, sure Bora. So first of all, a majority of those sales are brand new ports. And as the customer equipment is capable of taking GigE connections, many of them are willing and want to do that. We do have some swap outs, I would say that is less than 10% of incremental sales in the quarter where upgrades of existing corporate customers migrating from a FastE connection to a GigE connection. Today, we sell that GigE interface at about a 20% premium to a FastE interface. We think that differential will eventually disappear.And at some point in the future, when CPA of customers becomes ubiquitously able to take Gig interfaces, we think Cogent's standard product will become the Gig product today, not all customers can accept that. So rather than seeing ARPU decline on the corporate side, we'll just see the connection size increase. In fact, our average corporate utilization of a connection has fallen from about 18% to 12% in large part because of this proliferation of GigE interface.And then the final point, which is kind of a unique one to Cogent it's actually cheaper for us to sell a gig than it is an FE and the reason is, we do not need to put a media converter in the customer suite to step down the fiber connection to a copper connection. So all of our connections are delivered to the customer suite via fiber. But if the customer is still running a 100 Meg connection, we have to step that connection down from a Gig to 100 Megs and convert it from optical to copper. That's about a $300 piece of CPA that's typically charged in our customer install NRC and if the customer takes a gig, that $300 cost to Cogent disappears. So it's kind of a weird fact pattern that we want to actually sell more for the same price.

There are no further questions at this time.

Well, I would like to thank everyone for their attention and support. We're very pleased with our results. I look forward to seeing a number of investors at upcoming conferences. Take care and talk soon. Thanks. Bye-bye.

Ladies and gentlemen, this concludes today's conference. Thank you for your participation and have a wonderful day. You may now disconnect.