As the holidays loom large, so does the urge to look ahead and speculate on what will come in 2015 and beyond. With us today to opine about where the internet infrastructure industry is headed and why is Gillis Cashman, Managing Partner at the private equity firm M/C Partners. M/C Partners is very tightly focused on communications services and infrastructure, and Gillis Cashman himself sits on a half dozen boards of directors including those of recent IPO Zayo Group and Involta, his most recent foray into the space.
TR: Lately there’s been a lot of talk about moving data centers closer to the edge, what do you think is causing it?
GC: At a macro level what we're seeing is the architecture of the data center changing. To date the focus has been around maximizing power efficiencies and building large data centers in areas of low cost power, however, in order to maximize application performance and security, the key inhibitors to cloud adoption, the proximity of the data center to the end-user is critical. There is a common misconception that with cloud computing, servers and applications can be anywhere and accessed over the internet, however, in order to maximized application performance, servers and applications need to be closer to the edge and the customer.
TR: Do you think this is specific to certain verticals or more of a general shift?
GC: I do think it varies with different industry verticals. We're seeing a lot on the content side where guys like Netflix and Google really want to get close to the cable head ends and the edge and build redundancy in the network itself not so much the data center. We're also seeing it in the healthcare industry, which was our thesis behind Involta -- the first deal we closed in our new fund. They build Tier 3, enterprise-grade, highly redundant data centers in smaller markets, e.g. Marion IA, Akron OH, and Boise ID, where those kinds of facilities just didn't exist.
For large enterprises in those markets the only option had been to build it themselves. For example, there was an institution in Akron, Ohio that was thinking of building its own facility there. With Cleveland 40-50 miles away, when it came down to it the bandwidth costs of transmitting data there and back were cost prohibitive and the latency issues caused their application to degrade after 35 miles. There is a real need for this type of infrastructure in these smaller markets, and it really just does not exist today. When you dig into the demand dynamics in those markets, proximity is really important and it has to do with the applications these verticals are running.
TR: Why do you think this is just the beginning of a shift toward data centers at the edge?
GC: You have to think about what kind of network architecture is really needed. If Netflix or HBO had every one of their servers in one data center, and everyone was pinging Game of Thrones or whatever show, it would blow up the network. In order to distribute that sort of content, it has to be very well distributed with multiple points of presence out at the edge. Today's networks are getting extremely stressed with over-the-top internet video. But when you look at time spent viewing video online, it's still around 5-7%. From an infrastructure perspective that's a very scary stat when you think about the amount of disruption that level of penetration has caused, because you know that's going to increase to 50% with time. If we look at how the younger generation consumes video, we know that the infrastructure and the architecture have to change to support this kind of usage pattern or growth.
TR: Is this upgrade cycle similar to what we’ve seen before?
GC: What we are seeing with fiber today is very similar to what we saw with wireless and cable in the 80s and 90s. The big markets got built out, then there was a round of consolidation, and then over the next 15-20 years the smaller markets got built out. When you think about what happened over the last 15 years, there was a ton of fiber built out in the late 90s way ahead of demand, but it was really the longhaul and some small metro rings. The last mile was never really built. In '02 the music stopped and the industry went into a massive restructuring phase and over the course of the next 5-6 years there was very little investment in networks. I think the latest stat is that about 60% of buildings with 20 or more employees are still connected with copper facilities but it’s more pronounced in smaller markets. All of these buildings will eventually be connected with fiber. We are only in year 7 of a 25+ year upgrade cycle.
TR: When metro fiber went from nowhere to a hot new investment idea in just a few years, what changed?
GC: We have had all these advancements at the edge: faster enterprise services, laptops and smartphones. In the late ‘90’s, the network got way ahead of the edge when it comes to investment. But now the exact opposite is true, the edge is way ahead of the network. Whether it's the AT&T initial launch of the iPhone blowing up the network, or Apple releasing a new iOS platform and crashing New York Wi-Fi networks, or HBO Go crashing when Game of Thrones comes out, there's a new infrastructure need and it's at the edge of the network. We see a pretty long upgrade cycle ahead.
TR: Do you think there are still new opportunities out there in internet infrastructure for smaller players to take advantage of?
GC: What's interesting for guys like us who are mid-market investors is that for the enterprise it really comes down to a local business with local relationships, local rights-of-way, etc. It’s a misconception that Verizon and AT&T or even Zayo and the like have massive access to capital and can just solve this issue. It doesn't matter how much capital you raise, it is really about those local relationships and local networks you have in place. It's about scale at the local level, not at the national level. So there are plenty of opportunities in smaller markets on both the broadband and IT infrastructure side going forward. At a more macro level, the whole migration to the cloud is driving a lot of service ecosystems around specific applications and platforms that are creating new markets and opportunities.
TR: Do you think the big carrier neutral data center providers like Equinix are going to start looking more into the smaller markets too?
GC: I do think in the coming years you will likely see folks like Equinix move into the smaller markets. Right now they've got a lot of demand in the Tier 1 markets, but they'll want to set up points of presence in smaller markets over time. It's about point of failure and latency.
TR: There’s been so much M&A over the past few years in the fiber space, do you think the recent phase of consolidation is just about over?
GC: I think you've definitely seen the big wave of consolidation pass, and I don't see it being as active in the US. There will definitely be more consolidation, but I don't think you'll see the same number of deals. I do think you'll see another wave in the coming years as these smaller markets get more attention from the bigger guys and the guys who are operating in those smaller markets get to a scale big enough to appear on peoples' radar. If you look at the industry patterns, they tend to repeat themselves. There's an initial wave, and then another a number of years down the road. But for the next 6-12 months I think it will be relatively quiet in the states. Europe is a big question mark. While Zayo has obviously been doing some deals over there, each country has its own industry structure and economic issues, and it's not as definable. There is a lot of opportunity, but there are also a lot of assets that you wouldn't want to acquire as well.
TR: Do you expect to see much investment in longhaul fiber networks going forward?
GC: On specific longhaul routes, definitely, now that it’s been significantly consolidated. Local networks and the last mile are much more difficult to build than longhaul. And it also goes back to latency and the shortest path, and so there may be a route connecting two cities but if it's not the most efficient then it may not work for certain applications customers are trying to implement. So we're definitely seeing an uptick in longhaul activity. I think the positive competitive dynamic for the industry is that the number of speculative builds out there is next to none. They are always tied to contracted revenue. Even at significant demand, you've got a competitive structure that's been rationalized by consolidation.
TR: What about submarine cable networks, are those of greater interest to private equity investors yet?
GC: Over the last two years I've seen more business plans in that space than in the prior 15. I definitely get the sense that there is a need for more capacity. The issue I have there is that I've never seen a submarine deal where people made money on it. So from a private equity perspective, I'm not sure it's a rich vein for us. A lot of those deals are more strategic in nature with project financing where you have guys with very specific pain points they are looking to solve. But there have been a number of them, one going to Latin American and one, I believe, that even goes through the Arctic.
TR: How about on the regulatory front? How do you think the issues around network neutrality should play out, and how long do you think it will take?
GC: My view on net neutrality is that it comes back to internet video. There just is no business model around internet video for the carriers. When you have the dynamic where they're expected to invest billions of dollars in their network with no way to monetize it, I just don't know how you expect them to do it. From what I've seen to date with the FCC allowing deals between carriers and Netflix, that's a positive. They're concerned about the consumers, but someone has to end up paying for it. So there has to be a defined business model and value chain that makes sense for everyone. It's in its early stages, but those deals between Netflix and the carriers make sense to me and I think you'll see more of them. I think the content guys are going to be incented to form these kinds of relationships with the carriers. When you're only at 5% of the market it's not an issue, but at 50% they're going to want to have priority of service and secure delivery of their content. As the market develops, I think they'll be incented to work with the distribution guys and not go so much over the top. But there are a lot of constituents in that value chain and one thing I do know is that these things take lot longer than people think to play out.
TR: Thank you for talking with Telecom Ramblings!