Cogent Communications Holdings, Inc. (CCOI) Earnings Call Transcript & Summary

Great. Let's get started. Good afternoon, everybody, and welcome to our Cowen's 51st Annual TMT Conference. My name is Greg Williams. I cover the cable, satellite, telco and towers stocks here at TD Cowen. I'm here with Cogent, Dave Schaeffer, Founder and CEO. We have a 30-minute discussion today. With that, we'll just jump in. Dave, welcome, and thank you for joining us.

Hey, Greg. Thanks for hosting me. I'd like to thank Cowen for a great venue. I wish I could say I was here for all 51 years, but it has to be quite that long and most importantly I'd like to thank all the investors for their time today.

Great. So just to kick things off, let's get started with the Sprint acquisition. It's been a full month since you've officially closed on Sprint. To start the discussion, we maybe thought you can give us a general sense of how the integration is going and if there's anything faster or slower than you expected since we last spoke at earnings?

There's always puts and takes, but I would say, in general it's about as expected. Let me be a little more specific. I think the customer base is a little less pure enterprise, and there are a few more what I would consider general corporate customers. Secondly, the employee base has probably been a little stronger than I anticipated. And I would say the management and systems have been weaker than I had anticipated. But when we put it all together, I think the business is about in total, what we expected. And in terms of speed of integration, we have decided on an organizational structure, a set of processes, the product definition was done back at signing, and T-Mobile had done a very good job end of life-ing many of the gross margin negative products. Now because the acquisition closed a little earlier than anticipated, the revenue run rate was a bit higher than we expected. Burn rate was also slightly higher, but we still feel comfortable that within a year of closing, we'll be able to get that burn rate down to our 1-year target of $80 million negative EBITDA 2 years to 0. And by 4 years, $90 million of positive EBITDA. We feel quite comfortable that the cost elimination, coupled with the ability to preserve the customer base should allow us to do that.

Got it. And if we can talk a little more about the cost cutting. One of the big things you're doing is pruning a lot of the products and T-Mobile is doing that as well, going from 30 products down to 4. Can you help us with the process, what goes into that, the time line of this product rationalization?

Yes. So there were about 1,400 customers at signing. We anticipate about 1,200 of them will remain with us once this pruning exercise is complete. The average customer has a relationship with Sprint of greater than 30 years. The vast majority of their spend is either for MPLS or DIA products. The managed network services products were a wide variety of security and application management services. Those products along with SEP have been ended life. Those end-of-life notices were sent out almost immediately by T-Mobile at signing. Customers did have cracks. For those who are out of contract, those products were immediately terminated. Those that have contracts, we have an obligation, and we'll support those products until the end of contract, but we'll not renew them. Now for the DIA customers, we're migrating as many of those as possible to on-net, where practical, and we are increasing their port speeds without charging them additional monthly revenue. For those that are on MPLS, it's a 2-part transition. One, we are also increasing port speeds and moving them on-net, much is the same we're doing with the DIA products. But the second part is converting the back-end protocol from multi-packet label switching to virtual private line service. The reason for doing that is it's more stable, it's more scalable, and it's a product with a 20-year runway ahead of it versus MPLS, which is a product that is really end of life by all of the equipment vendors.

Right. It's interesting that you're migrating folks from MPLS to VPLS and couple of years ago, it was MPLS was secure, controllable and had more control of the network, can you say that now that you're migrating because the technology come to a point where it's pretty much indifferent on the front end and just different in the back end?

I think the quality of service to the customer is equal or better than it was with an MPLS platform. It does not require a piece of customer premise equipment. It allows for a very flexible on-the-fly reconfigurations and it allows for full line rate throughput. The other competing technology, which does have some market traction, but today is maybe not quite ready for prime time, is the SD-WAN. And the approach of the 2 technologies, VPLS and SD-WAN are different, the net results are very similar. In the case of VPLS the service provider's network is taking in packets, encapsulating them or putting a wrapper around them and then transmitting them over the Internet and unpacking those envelopes. In the case of SD-WAN, the service provider is taking individual packets, encrypting them with IPsec to 128th power encryption keys and then decrypting on the other end. The downside of that is it requires a tremendous amount of CPUs. And as a result, the actual throughput in is much lower than the line rate. Compare both of those approaches and using the public Internet to MPLS. MPLS takes a layer 3 network, like the Internet and builds a pseudo circuit, a false circuit across the network. It's kind of tricking a packet network and thinking it is a circuit switch network. The amount of overhead in doing that is just prohibitive, and the amount of network management that's necessary to keep that operating well is just prohibitive. So MPLS just becomes an unscalable technology.

Right. And you mentioned earlier that customer base might be less enterprise and more corporate. Can you help contextualize what that means and what that means for Cogent?

Yes. So as we look at the roughly 1,200 remaining customers. As I said, there were 1,400 customers at acquisition due to the fact that we're end-of-lifing these noncore products. We expect about 200 of the 1,400 customers to go away based on the fact that the majority of their spend were in these noncore products. Of those remaining 1,200 customers, about 450 of them are truly enterprise. Now they account for almost 80% of the revenue. When we did our due diligence, we were shown data on all 1,400 customers, but the names are redacted, but all but the top 5. And that was for FTC purposes because the transaction did not close for some reason, we could not have that competitive intelligence. So we were flying somewhat bind. On May 1, when the deal closed, we obviously unmasked all of the roughly 1,300 something of the 1,400 customers that were still with the company and looked at them in depth, we realized that while it was represented to us that it was mostly enterprise revenue, and that is true, by customer account, there were a lot of companies that were sub-$5 billion of revenue, sub-1,000 employees and wouldn't fit in the kind of enterprise definition that we would use. So we'll melt those customers into the Cogent corporate organization, and we will take a handful of enterprise customers that Cogent had sold to over the years and consolidate that with the Sprint sales organization that will be focused entirely on maintaining and possibly growing this enterprise base. But the reality is these customers are very sticky, very hard to attract new customers, and there's not a tremendous amount of growth opportunity. You can see that across the entire enterprise landscape.

Right. I want to talk about the waves opportunity from the Sprint GMG acquisition. I think it's an underappreciated opportunity. You've noted you can ramp up to a $500 million revenue business in 7 years from nearly nothing today. So help us on how you expect to dominate the win share and achieve $500 million in -- which is essentially a core of the market?

Yes. So Cogent's business prior to the acquisition was selling 3 products: Internet access, which is 81% of classic Cogent's revenues; VPN services, off of a combination of VPLS and SD-WAN. That's about 16% of revenues and 3% of revenues came from colocation space and power in the 55 data centers that classic Cogent operated. When we acquired Sprint, we acquired 3 things, an enterprise customer base that we've just finished talking about. We acquired 1.6 million square feet of central office space that is across 500 buildings. 45 of which are appropriate to convert to data centers with approximately 150 megawatts of power. So we are in the process of connecting those buildings back to the Cogent network and removing debt equipment from those facilities and are going to sell space and power, both on a retail and wholesale basis. And then finally, the primary reason for doing a transaction was the ability to secure fiber on the 19,000 route mile intercity network and 1,100 route mile metropolitan network. 90% of those routes are fully unique. Those routes, however, had little marketable value because the Sprint network only connected to 23 data centers in the U.S. Cogent's metropolitan network connected to 800 data centers. By physically interconnecting the networks at multiple points, we expand the total addressable market to all 800 facilities. The Sprint network was not a roll-up. It was organically built by 1 organization with very accurate engineering data. In fact, we can auto generate a map with every order that will show the exact routing with 1 meter accuracy, left or right, from the train right of way with each order. No other carrier has that level of inventory management. As we think about our competitive advantages, it's the uniqueness of routes. It's the ability to have accurate engineering data. It's the ubiquity of endpoints. We will have 800 carrier-neutral data centers in North America, where we could sell wavelengths. Today, we have 200, up from 23 just in the time that we were able to preposition prior to closing in the 1 month since closing. And then we will have the ability to price aggressively. And we already have a sales force that covers this market segment. Depending on how you measure it, Cogent is either the largest or second largest carrier of Internet traffic in the world. You can validate that by talking to our competitors as much as hearing it from me. And we have about 250 salespeople that focus on that market. That is the same market that buys wavelength services. There are about 200 material wavelength buyers. Cogent already has a commercial relationship with 3/4 of those or 150 of those 200 are buying transit today from Cogent. We intend with the ubiquity of reach and the uniqueness of routes, coupled with aggressive pricing to go from an $800 million -- or basically an $8 million run rate annualized about $800,000 a month. So $150 million to a $500 million annual run rate in 7 years. There's about a $2 billion addressable market. We think a 25% market share, which is what we've already achieved in transit is reasonable. And the fact that we already have a sales force and we have brand credibility puts us in a very different position than Cogent was 20 years ago.

Right. And it's a roughly $2 billion market for U.S. waves. It's dominated by Lumen and Zayo. Do you expect to go after those customers or the other half, if you will, because the question I ask is when you buy waves from Lumen, you're actually buying usually solution sets. How do you pick apart just the wave product from a Lumen or Zayo, steal the share.

So I actually disagree with the premise of your question. I think the customers for waves are buying just wave, not solutions. So if you think about the customer base, there are 3 major categories: hyperscale and content-generating businesses, regional access networks. And both of those customer sets, there are no solutions, so just buying point-to-point pipes. The third segment are very large corporate customers. There's probably 50 corporate customers who can justify owning a wave network. Waves are more expensive than using the Internet. They have some advantages. The route is deterministic, meaning you know exactly how the packets are going to travel. Two, the route is secure. But three, the route is unprotected. And fourth, it is more expensive. So in building the network with waves, a corporate customer will typically look for diverse paths to protect every span. And again there are companies that value these other attributes. For hyperscalers, it's particularly desirable because they're doing large bio transfers for data replication. And the Internet, while it's great for small pockets, it's not really good for sending huge amounts of data. So if you're trying to move all of the data from one Google data center to another, Internet is probably not the best way to do that.

Right. And that's going over public routes versus the wave stays in your network?

The wave stay -- well, the bit stay within that wave. And the bits are kept coherent as opposed to broken in a different pockets that are then sent in different paths and reassemble at the other end.

And you've noted in the past you can provision a lot faster, too, and ramp these customers up?

So one of the advantages that we've established in the transient market is our ability to provision quickly. So we guarantee customers a 17-day on-net provisioning window. We actually achieved 9 days. Today, in the wave market, the market is somewhere between 90 and 180 or more for provisioning. We initially are guaranteeing 90. We believe that within 2 years, as we fully standardize the network and we add all of those carrier neutrals, we will bring that guarantee down to 17 days to match our current transit agreement.

Right. So 4 to 5 months down to 17 days.