Bank of America. This is our 2022 Leveraged Finance Conference. Finally back in person in Boca Raton. We're thrilled. We're also thrilled to have Cogent Communications with us. Dave Schaeffer, the company's CEO. Dave, thanks so much for being with us.
Hey, Ana. Thanks for hosting me. I'd like to thank Bank of America for a great venue. I'd like to thank investors for taking time out of their day to hear a little bit about us.
Great. A lot to cover. I thought the way we would structure our conversation today, Dave, is maybe kind of in maybe equal portions. First, topics related to what we call Cogent Classic now. Then you've got the pending acquisition of the Sprint wireline network, so a lot of questions and topics related to that. Then finally, because this is a debt conference, we wanna touch on the company's capital structure. Does that sound like a plan?
Sounds great to me.
Okay, great. Starting with Cogent Classic. The demand environment, the churn environment. Starting with the corporate business. After the pandemic-related headwinds, that resulted in negative growth, I think really for the first time in your corporate business.
Corporate history.
Yeah. You were able to get it back to flat sequential revenue performance. Among the trends that you're been citing is slow back to office momentum and then the heightened corporate vacancies in your office building portfolio. What is the current vacancy rate in your office building portfolio versus a historical norm? Do you think that as leases come up, there's still gonna be a persistent headwind for the foreseeable future on these vacancies?
Cogent today offers its corporate services in about 1,825 skyscrapers in North America, approximately 1 billion sq ft of net rentable office space. That's about 11% of the inventory in North America is directly on net for Cogent. The vacancy rate in that footprint pre-pandemic was about 6%. It skyrocketed during the pandemic to 18%. It stabilized, and today it's at 17.7%, so a slight improvement. We are seeing a pickup in leasing activities, a decline in rental rates. Most of the new leases tend to be for smaller offices on shorter terms at lower rates. I think much of the equity that owns the buildings that we're in may be in trouble, but the buildings themselves will probably fare well. With previous economic downturns, what we have seen is a flight to quality.
Tenants typically want to move from B and C buildings into the most prestigious buildings in a city. You know, not everybody can afford to be in a 9 West 57th Street or a Hudson Yards under normal conditions. When there's a condition like the pandemic, those rents fall and people that are in midtown wanna migrate to these better buildings, and that's true across the country. A second way to look at this is by card swipes. That's basically the number of employees that are entering the work space on a daily basis. If you use kind of February of 2020 as a benchmark, we had fallen to only 2% at the worst of the pandemic. We have returned today, and we're at about 50% of card swipes compared to pre-pandemic levels.
That means the number of employee days in the office is half of what it was pre-pandemic. It's very geographically uneven. In the South and the West, it's over 90%. New York's running at about the national average, at about 50%. If you look at the most depressed markets, Seattle and San Francisco, they're still below 30% of employee workdays in the office.
Okay. As you're getting these new tenants moving into these buildings, what is your win rate like? With the smaller footprint, how cost-conscious are they? You know, what does that mean for your pricing?
Yeah. Cogent's corporate product is priced at parity with our competitors, which are typically the incumbent and one other competitive carrier. We differentiate ourselves by offering a install time that's nine times faster on average than our competitors, a service once delivered that is three times more reliable, and then finally, 30-60 times the actual throughput. We win about 50% of the proposals that we make. Today, we are about 15% penetrated in our corporate on net footprint. As the average office size shrinks, that actually is a good leading indicator for us because we sell our corporate product on a fixed connection basis.
If the building returns to historical occupancy levels, there will be more discrete businesses in the building and therefore a larger TAM. We have been able to go from, you know, kind of -8% growth in our corporate business to flat year-over-year growth. We anticipate that to continue to improve and get back to that roughly 11% year-over-year growth rate, albeit it's slower and less even than we had originally anticipated.
Okay. Two other things that impact your ability to sell is, one, your sales force. And then obviously, a prevalent topic of this conference is the potential for macroeconomic headwinds or a recessionary environment. Labor's been an issue and labor productivity is always an issue for you, given that you have an outbound telesales, corporate sales force. Where are you now with having the labor that you need to get back to those historical corporate rates? If you can just comment on what you're seeing macro-wise.
Yeah. We continued to hire during the pandemic. We sent our employees home, and we were training them remotely. Pre-pandemic, we were averaging 5% per month turnover in our sales force. It is an outbound telesales organization. We have 80 teams that are located in 45 offices around the world who spend over 90% of their customer interactions either over the telephone or through email. 90% of our sales occur with the salesperson never physically meeting the customer. During the pandemic, we saw our sales force turnover rate increase from 5% a month to 8.7%. When we brought our sales teams back to the office, we have seen that turnover rate decline. We're today at about 5.5%, still 10% worse than we were pre-pandemic, but a significant improvement.
Our sales force productivity is measured by orders installed per rep per month. Had averaged 5.2 orders per month. It declined during the pandemic to 4.2. It's back to 4.7. We are seeing improvements in rep productivity now that we have our employees back in the office where they can learn from their colleagues, have direct managerial supervision, and also have the ability to have in-person training. In terms of applicants, we have never suffered from a lack of applicants. Pre-pandemic, we were averaging 13.5 applicants for each job opening. That actually has improved, and today we're seeing about 15 applicants for each job. Our initial starting salary is $80,000 a year. A typical rep's probably 28 years old, you know, soft degree from a state university.
To be able to telemarket and make $80,000 is pretty attractive. The challenge is it's a job that requires 100 outbounds a day, and very few people can do that over the long run. Many people think they can do it until they actually do it. Anyone who's ever had a telemarketing job can, I think, attest to how hard it is in reality.
Okay. On the NetCentric side, demand really benefited during the pandemic. Are there any signs of that slowing? If you could just talk about the pricing environment there and the competitive environment.
Yeah. 97% of Cogent's traffic, 43% of its revenues come from selling wholesale bandwidth on a meter basis in data centers. That is heavily international. 55% of that business comes from outside of the U.S. There are two major customer groups. 7,750 regional access networks buy their upstream from Cogent, and we sell to about 5,000 content generators. What we have seen is faster growth outside of the U.S. and a greater percentage of traffic that we're getting paid by both the sender and receiver. Today, 55% of our NetCentric traffic and revenues come outside of the U.S. 73% of our traffic, we're getting paid by both the content and the access network concurrently. The average price per megabit has declined for the past 25 years at a compounded rate of 23% per year.
It's effectively at that same rate of decline today. It was 21% last quarter, a slight improvement. It oscillates around that 23%. Offsetting that has typically been traffic volume growth of about 25% per year. The total addressable market for transitThe bulk sale of internet connectivity is about a $1.5 billion TAM. Today, Cogent has 24% market share by bits. We have approximately 15% market share by dollars. We continue to gain market share. We guarantee to undercut our competitors by 50%, and for that reason, we have been growing by gaining market share. During the pandemic, we saw a significant acceleration in the adoption of streaming. In 2019, approximately 18% of video was distributed via streaming, with 82% of video being distributed linearly. Today, 44% of all video in the world is streamed.
That number has continued to increase, and as a result, we have seen an increase in our NetCentric business. We went into the pandemic with our NetCentric business growing 3% a year. Last quarter, it grew 16.5% year-over-year.
Okay. I think it's a good time, we're about halfway through, to shift to Sprint. Sprint wireline. Fascinating transaction that you announced a few months back. I think you talked about it. You had a public call and addressed it on your earnings call. I think the facts are out there. You're basically buying actually being paid to take the Sprint wireline network from T-Mobile. Paying $1 just to say you bought it, they're paying you $700 million, $350 million of which I think is in the first year.
What you've cited, just in to recap, is that at closing, which should be by the end of 2023, you expect it to have $450 million of annualized revenue, and to have negative EBITDA of $180 million and probably need about $30 million of CapEx. You know, right there, it's $200 million-ish cash burn day one. Simple math on just the EBITDA is that this is a network that has $450 million of revenue, but a cost base of $630 million. If you could enlighten us on, you know, what happened, and you know, what is in this cost base and what you can do with it.
Let's start with the underlying network and the customer base. Sprint built the first nationwide fiber optic network in the 1980s. It was built with SMF-28 fiber that actually turned out to be optimal fiber for today's high throughput coherent technology. It was built primarily along railroad right of way, direct buried in armored cable. That network supported $40 billion in long distance revenue. There were 70,000 employees in that business. It was sold, or potentially sold, to MCI for $129 billion in 2002. That deal was blocked by regulators. That deal was blocked, Sprint's management focused exclusively on becoming a wireless company, and for 20 years had underinvested in that wireline business. It shrunk from 70,000 employees to 1,320.
It went from $40 billion in revenue to $560 million in revenue. What we are acquiring are really two things, a customer base of 1,200 large enterprise customers. Today, there are 1,400 customers. There are 28 products that are being sold. The majority of those products are gross margin negative. During the period between signing and closing, T-Mobile is end of lifing 24 of those products. We expect about 200 of the 1,400 customers to go away, and the remaining revenue to be about $450 million in the four products that we will support going forward. Those 1,200 customers are very large enterprise customers, very different than what Cogent has today, and we do not anticipate growing those customers.
We also believe that by increasing the bandwidth per connection by 10x, by bringing those customers on net, we can retain that $450 million of revenue. The second thing that we're acquiring is effectively an empty network. What will happen on day one, we will transition all of the Sprint traffic onto the Cogent network. Cogent today carries about 1 exabyte a day of traffic. Sprint carries about 10 petabytes, about 1%. They would look like a small or mid-sized customer to Cogent. We will put that onto the Cogent network, making the entire Sprint network empty or dark. We will repurpose one of the pairs of fibers to sell wavelength services, optical transport services. That is a product that Cogent does not sell today. It is a $2 billion addressable market dominated by Lumen and Zayo.
We will interconnect that network to our 18,000-mile metropolitan network and bring 800 carrier- neutrals in North America on net day one. We think within 5 years, we can eventually capture 25% of that addressable market with our aggressive pricing and the repurposing of the network. We will light a second pair of fibers and migrate the entire Cogent IP network onto that pair of fibers. All of the remaining fibers will be available for sale as dark fiber. We will also take the 1.3 million sq ft of fee simple data center space that Sprint had built to house its telephone central offices and repurpose that for colo. The physical asset that we're acquiring is a value-added asset that is being repurposed. We are paid $700 million by T-Mobile to effect this transition. We will reduce the headcount.
Some of that is going on now in the pre-closing phase. Some additional will probably go on post-closing. We also will rationalize the networks by having Sprint abandon its leased international network, move on to Cogent's own network, bringing customers on net. Today, 93% of Sprint's end users are off-net, where they're buying circuits from a telco. At Cogent, 75% of our customers are on-net, and only 25% are off-net. We will utilize our footprint to reduce those local access costs. We expect that the burn rate in year 1 will go from $180 million negative EBITDA to $80 million. By year two, it will go to zero, and it will be cash flow positive by year three. Within five years of closing, the revenue of the combined company will be $1.5 billion.
The combined company will have a growth rate of 5%-7%, and it will have EBITDA margins in the low to mid-30s%.
Okay. A couple of questions on that front. One, I mean, obviously there's the synergy factor that is, you know, driving the transaction. It sounds a little complicated, so I think there is a concern about execution risk. Then secondly, how did Cogent end up with this asset? There are other players out there, I think you mentioned two of them, that presumably could have had the same sort of synergy economics. There's that question, and then two, I mean, if you're worried about something going wrong in the execution, where is there the greatest risk?
First of all, Cogent had affected 13 acquisitions in its formation, buying large companies like PSINet, Allied Riser, FirstMark Communications, Carrier1, Verio, and integrating those assets into the original Cogent business plan. Between December of 2004 and August of 2022, we looked at over 800 transactions, and this is the only one that we did. We actually looked at this asset twice previously, once when SoftBank announced its acquisition, once right after T-Mobile acquired SoftBank's stake in Sprint. In both cases, we were unable to reach economic terms that made sense to both parties. When T-Mobile eventually decided that this asset was non-core, they ran a process. There were about 20-something companies that took a look. About 10 did hard due diligence. T-Mobile had three major objectives. One, no brand damage to T-Mobile.
They spend $3 billion a year on branding and did not want any negative brand connotations from shutting this business down or mistreating customers. Two, they wanted no regulatory issues as they had made many commitments to regulators, and they did not want to jeopardize their wireless acquisition. Then third, they wanted to look, you know, relatively smart in doing the deal. We were in a competitive process. We were actually not the best economics. There was another group that was willing to take a smaller payment from T-Mobile than our $700 million payment that was settled on. We convinced T-Mobile that our track record on our ability to execute gave us an advantage, and they agreed with that. Now, what can go wrong? You know, as in any business, it's a series of tactical executions, and it starts at the customer and the circuit level.
You know, we need to maintain the existing customer relationships. We need to migrate those services to our own facilities on a circuit by circuit basis, eliminating those off-net expenses. We need to rationalize the employee base for a more modern product set. You know, Cogent going forward is gonna have four products: Internet access, VPN services, colocation, and wavelength services. That's very different than any other telco, and it's our product focus, our simplicity, and our sales execution that I think will allow us to succeed. You know, Sprint only had 60 salespeople out of 1,320 employees. Cogent has 750 in sales out of 1,050. We are very much focused on growing revenues.
Okay. From a structure standpoint, you're putting this asset in sort of a non-recourse subsidiary with regard to the existing debt holder group. I think from a consolidated perspective, it all rolls up, but there is a ring fencing of this asset from a kind of restricted group perspective as debt holders like to talk about. What is the rationale with regard to that? At what point might you merge the two back together?
Our biggest concern was how we can account for the $700 million that we get from T-Mobile. We are providing them transit services. They have publicly said, including their public statement at the Bank of America Equity Conference several months ago, that they do not intend to use the services. We would like to count that as revenue, but are not certain that Ernst & Young, our auditors, will support that. We think they will, but we don't know that for a fact. It made sense for us to isolate the burn and the payment stream into a unrestricted subsidiary that is a sister subsidiary to our borrower. Cogent has a HoldCo OpCo structure. There is no debt at the holding company level. All of the debt exists within Cogent Group, which would be the Cogent Classic business.
For our bondholders, they are not taking the risk that some of that payment is not counted as revenue. We have a interest coverage trap on our restricted payments. We needed to be able to access that accumulated cash for dividends at the holding company level, and can do so under this structure. The bondholders will actually be equal or better off every day. The first thing that'll happen is the bondholders will see the payments that were going to Lumen for maintenance on our IRU go down when they are now paying to Cogent. Second, they will see Cogent Group deriving revenue from Cogent Infrastructure as it leases them infrastructure outside of the U.S. Third, as we migrate the customers onto the Cogent IP network and Cogent product set, they will all fall under Group, increasing Group's revenues.
In the Infrastructure Subsidiary, the physical asset, the $700 million payment from Sprint and the sale of infrastructure assets will exist outside of the credit group. The cash burning entity will be outside of the credit group, and I think the bondholders from day one will be better off. Over time, as the financial profile of the Infrastructure Subsidiary improves due to wavelength sales, it may either become a standalone borrower on its own based on its positive cash flow, or we would then have the ability to merge that together and consolidate it. That decision has not yet been made.
Okay. I hate to say it, this time flies, and we're out of time already. Much to cover. Dave, thanks so much for being with you. It's fascinating, and we'll be watching very closely your progress.
Hey, thanks, Ana. Thank you all very much.