Equinix, Inc. (EQIX) Earnings Call Transcript & Summary

All right. Well, good afternoon, everyone. Eric Luebchow here of Wells Fargo and really pleased to be joined today by Equinix. We have Jim Poole, who's the VP of Business Development; and Katie Morgan, who is VP of Investor Relations. Thank you both for joining us.

Thanks for having us.

Thanks for having us, Eric. And just a quick disclosure before we kick this off, but we may be making some forward-looking statements today. So please check our SEC filings for those disclosures.

All right. Great. So I'll kick it off to either one of you for the first question. Just kind of recapping your recent Q3 earnings report, you guys talked about record gross and net new bookings in Q3. It's been a really strong year despite kind of an uncertain macro environment more broadly. So maybe you could talk about which demand verticals are really working right now across service providers, across broad enterprise and maybe give us a little bit of insight into where you're really seeing a lot of the data center adoption today?

Sure. So I can start, and then you can jump in. So I guess, I'd break it down into kind of 2 different views. One view would be sort of regional. So if you were to look at specific areas of the world, some are hotter than others, right? So if you were in the U.S., that would be New York and Silicon Valley. If you were in Europe, that's places like Dublin or Frankfurt. If you're in Asia, Japan is a very hot market. So that's kind of one lens you could take. The other lens you could take from a vertical perspective would be there's continuing on the service provider side a massive amount of investment that's going on between the network service providers and the cloud service providers as far as being able to move traditional home data center workloads over. So we see a lot of benefit from that on the interconnect side of what we do, and the systems that are necessary to interface with one another between those counter-parties land in our facilities. So that's a very durable, strong piece of the business. More generally against most other verticals, it's a question of where they are in their sort of digital transformation. So you have sort of cloud-native sorts of traditional companies born in the cloud. So think of a SaaS company, right? That's one kind of customer. Increasingly, for us, enterprise is the fastest growing part of what we do. And the reality of that is the fact that enterprises, for the most part, there's a couple of truisms you can use. One is that they all use more than one cloud provider. Not everybody ever uses one. They use more than one network provider. So that's two. And then three, something always stays private in what they do. So even if 80% of their workloads go to the cloud, something stays resident in the data center and wants to be cloud-proximate, right? And so if you looked at manufacturing or health care or financial services, usually traditionally brick-and-mortar enterprises who are sort of still going through that digital transformation process, those are the strongest areas we're seeing right now.

And I guess, just kind of related to that, as we think about broader hybrid cloud adoption, I mean, we've been talking about it now for years. And it doesn't seem like anything has really slowed down. So where do you think we are in terms of enterprises moving things from like an on-premise architecture to adopting some type of hybrid cloud, which could involve some public cloud, some private cloud, perhaps some other private network services adopting a hybrid architecture, where are enterprises on that journey and perhaps it differs by different industry segments as well?

Yes. I mean, overall, definitely, I think what you saw sort of -- if you backed up 10 years, what was common is cloud started becoming a hot thing, a lot of enterprises would start to just willy-nilly export workloads into public cloud. And then the reality was they found out not everything works well in public cloud. And so we call it repatriation. I call it precipitation just because it sounds funny. But the reality is that is what customers have to do. Now there wasn't a whole lot of industry knowledge for an enterprise to tap into that says, "Hey, you probably shouldn't have done that. You shouldn't have tried to go all to one side of the boat. You probably should have landed somewhere in the middle." So nowadays, the nice thing is whether they're the network service providers, the cloud service providers, the MSPs that we do business with, for the most part, all of the kind of vendors you would traditionally go to as an enterprise to implement a hybrid multi-cloud architecture, it's a very well understood thing that, that is a combination of different kinds of assets and that the interconnection of those assets is critical to the success of any of those sorts of projects. So that would be the easiest way to think about it.

And in terms of repatriation, I mean, we've heard from the hyperscalers who reported earnings recently, there are some concerns around slowing revenue growth, albeit still very impressive levels of revenue growth off of a very large base that certain customers are tightening the belts a little bit in terms of cloud spending, ingress/egress costs that have gotten a little bit out of control in terms of pushing and pulling data into and out of the cloud. Do you see any sign of that happening more broadly in terms of repatriation?

Not so much in our case, and I'll explain why, which is, it goes back to the fact that our cost to the end customer is a fixed cost regardless of how much they use it or what they use it for, right? So I'll say, customer could buy 4 racks and a bunch of capacity from us, and that's $10,000 a month, whether they run 1 meg of traffic through that infrastructure or they run 200 gigs of traffic through that infrastructure. And for them, it's more the fact that without those physical interconnect capabilities existing, so the deployment of the base assets, there is no utilization, right, to tune down. So it tends to have more of an impact, in our view, for very small customers who may have a project in one cloud, definitely could see where that could be impacted as opposed to our customer base, which tends to be much more on the service provider and multinational enterprise side. And so what they're trying to do with us is usually one of 2 things. They're eliminating a bunch of legacy sort of cost involved with being in their own data center or trying to optimize the economic equation between public and private. And so they're -- they have to be cloud-proximate. So the net of that is we are saving them money. But the other side of that is almost everybody is making their traditional services easier to consume by making them digitally available. And the biggest customers of those services for a lot of our customers tend to be other companies that sit in the same data center. So we are the largest aggregator of demand for a lot of people in the -- so it's -- you get this sort of network flywheel effect between companies doing businesses with -- business with each other in the facility. And that's very different than, say, a small company doing business with a hyper.

Got you. Got you. All right. That's helpful color. So Jim, I understand one of your areas of interest is edge compute. And Katie mentioned that you were an expert. So I wanted to maybe ask you about the word edge, which I often think is kind of overused by and misunderstood by a lot of people, especially sitting in my seat. And I've always thought Equinix is really kind of a leading edge provider today because you cover the vast majority of most of your markets with very low latencies. But maybe you could talk about some of the new edge use cases you guys are seeing in your portfolio, and then we can explore that topic a little further.

Sure, sure. And I love your logic because I use the exact same words when I talk to customers, which is my view of Equinix is we are the state-of-the-art of the edge as it exists today in the world. There is no one you can go to who has a more distributed footprint with access to as many clouds and networks as we do. So that's sort of table stakes. Now we are definitely mostly a Tier 1 market construct. So we tend to be in very, very large cities. And so when people ask this edge question, it's this thesis that says, hey, does the market develop beyond the current construct of being in these sort of large metros. And certainly, on the one extreme, you can go all the way down to something like a tower, right, which has got very little at it. Or you back up to something like us, which is what I like to call is the aggregated edge, right? So this idea that says in these locations, how do I get multiple networks, multiple service providers to build my supply chain around whatever it is I'm trying to do. And the reality of that is, that is a very difficult thing to replicate beyond a certain level, right? And so the example I'll use with use cases that we're dealing with would be something like, let's take GPUs, right, so hot thing these days. If you were looking at like a security application in a retail store using GPUs, traditionally, the way you would have done it would be you'd put a GPU on a camera. Each of those cameras was $750, and you'd have to put 100 of them in the building. And the reality of that was it was a wired solution. It was very expensive because you couldn't put that number of devices on a Wi-Fi network, right? So now with 5G and other things starting to happen, you can put $25 cameras, 100, 200 of them in a store and you can park a whole bunch of GPUs in a downtown multi-tenant data center like Equinix and get cloud economics against that infrastructure. So suddenly take the restocking, security, walk out, pick and go sorts of applications. All of those are computer vision applications against video streams. And so if you can aggregate the GPU portion of that, you can then extend the market applicability of that application to way more companies than traditionally would have been able -- what it would have been a very large company kind of a product. So that's just one good example.

Yes, interesting. I guess because you kind of brought it up in your response, I mean, I think the inevitable question with new use cases like 5G that are emerging, if we get to a place where people require sub-20 millisecond latencies, would you see more compute getting pushed into -- on one hand, you could see secondary, tertiary markets. And then kind of the other extreme is, of course, the tower site where American Tower just made a big bet on that themselves. Is that something that you guys are watching? Is it something that maybe you guys could take advantage of at some point if there was an opportunity there? Or is it just not -- or the economics, the business model not developed enough for you to really have a lot of conviction?

So I -- so one is we are -- we do watch it. So that's without a doubt. But I think it's important to give some context on sort of how we think about the problem. So as an example, take the United States. 80% of the U.S. population lives within 10 milliseconds of an Equinix facility, right? But the existing 4G network, average latency across the network is 40 to 50 milliseconds, right? So why is that? Well, all the previous generations of wireless were a consumer service. All the traffic got dumped in the Internet. And there was no target latency that you were trying to achieve because voice is very tolerant, right? You can do 60 milliseconds just fine, and you wouldn't even notice. So that's one thing. So what's happening? As the operators have invested in spectrum and radios, the first thing they did was of course turn it on, so you get a nice icon on your phone. And you think suddenly the world is a better place, but you don't notice any difference. And there's a reason for that. And the reason is they're still using the 4G core with these fixed locations where all the traffic dumps into the public Internet. So what we are doing now, our operator conversations are up around the disaggregation phase of 5G, which is to say, when can I take traffic off a specific tower and route it to a specific application hosting instance with deterministic latency and performance, which 5G is the first wireless standard that's capable of doing that over the radio interface. The gating factor is really a software function that needs to live at our facility, right? It -- all of the other pieces already exist, all the fiber that comes into our facility, all the transport infrastructure that the carriers have to deploy. So one of the things that I say with very full confidence to the operators is, if you are going to start unlocking business customer value against 5G, where is the first place you are going to disaggregate your network? It's not going to be in the same place the same way that you're always doing it. You have to start accounting for the largest pools of private IT infrastructure and public IT gateways that exist, which would be us. So we're now finally getting into that phase. And so you'll start to, I believe, see more of that. Now once you have that, to your question then is, okay, if I can get 10 milliseconds for 80% of the population, once I do that, well, then I've learned enough that says, okay, when would I then want to go lower? And I think as an industry right now, there is no good use case for sub-10 right? It's -- no one understands what that would be or at least how they would make the money to do it. But our argument would be, hey, we're in it with you already. We'll learn a lot together on this. And then when we understand what the economics of it are, then we can make the argument that says, okay, we need to go somewhere else. Because the problem is not a technology problem. The problem is an ecosystem problem. And for us, we've trained our customer base to sort of do the land-and-expand thing, right? They use our facilities, and then they push workloads into other facilities as they need to be more proximate. That behavior is not going to change. All that's going to happen is you're suddenly going to have a 5G wireless access capability hanging off the same infrastructure. And so the nearest-term monetization of 5G from a business perspective, my prediction would be, would be between us and the operators and the cloud providers just monetizing what already exists.

Interesting. Yes, that's a good perspective, especially useful since we are at -- been part of TMT telecom conference. So thanks for that. I'll divert a question to Katie now. So I wanted to talk about maybe some of your performance on churn metrics. I think one of the themes from your recent earnings call was that churn has continued to trend lower and lower. You were sub-2% the past quarter. I think some of the acquired properties that you've had historically like Verizon have certainly -- that had some embedded churn in them have started to perform better and better. So maybe you could talk about what you think is working, how sustainable it is that you can be at these lower levels as we look forward.

Yes. Thanks, Eric. As you noted, our churn metric has been trending lower. If you look over the trailing, call it, 8 or 10 quarters, it has trended lower. And it is partially attributable to some of the Verizon churn, we knew that going into that acquisition. It stuck around a little bit longer than we were expecting. But that has really worked its way through the system. And you've seen that within the Americas utilization rate. The Americas utilization rate has nicely stepped up from, call it, low 70s to about 80% utilized as of quarter-end. And it's really a reflection of just putting the right customer with the right application into the right footprint and being very selective in the customers that we're targeting for our facilities. We're not targeting every type of deployment out in the marketplace. It's again just putting the right customer, right application into the right footprint.

Okay. And that's helpful color. And I think one of the other themes that we've tried to explore with a lot of companies looking into next year is how recession-proof is any individual company. And I think Equinix certainly, given your history and how you performed during the great financial crisis, proved to be fairly recession-resistant. So maybe you could talk about conversations you've had with your customers in terms of, is there any sign maybe in small or medium businesses, which you don't have a lot of exposure to, that potentially could be pulling back spending or trying to rationalize their data center footprint to kind of tighten up in a very volatile macro environment.

Sure. I'm happy to start, and then you can add on, Jim. I'd say, overall, when you look over the course of our history, we just had our 79th consecutive quarter of top line revenue growth. So as you look over the course of our history as a company, you can't pick out the various capital market cycles or economic cycles that we have operated against. I'd say we sit in a very unique position where customers can really come to us for really one of 2 reasons. Number one, we can help them drive incremental revenue into their business and fuel their top line revenue growth. Or on the flip side of the equation, we can help them manage their total cost of ownership and reduce their expenses. So we really sit in a unique position. And again, focusing within our customers, targeting -- there's critical applications that can either be revenue-generating or revenue-facing. So we feel like we're well positioned. Jim, anything you want to add on?

Yes. I mean just to give more color to that because I've had these conversations with the big service provider customers that we do business with that says, look, if you were just to -- say you're a telco. If you were to stack rank every lit building you have and then you did an ROI analysis against the infrastructure you deployed in those buildings, where do you make the most money? It's always us. We are at the top in every metro we do business with them. So if you have to do rationalization as a company, you don't start at where you're making all your money, saving all your costs. You start at the other end. So we see that as the demand for what we do is highly inelastic because it's sort of a foundational fixed infrastructure, this is what you pay to play the game, right? If you don't do this, then you're not actually playing the game, you're doing something else. So from that perspective, we feel very strong. That's the reason we feel like we keep seeing the strong demand. We do our own -- I used to be in the marketing organization. And so I did some of this analysis. We do propensity to buy analysis of the enterprise market based on the behavior of the existing customer base that we have and thousands of databases that we query. And the reality is we have 10,000 customers, 6,000 of them are enterprises. However, our target market that fits our propensity models on a global basis is 400,000 enterprises. So when you have this massive shift of private on-premise infrastructure to cloud with some amount dropping out into the multi-tenant data center space and the market is only 40% to 60% of the way through, just the tailwind blowing in your direction is so massive that even a slight deceleration of the overall macro doesn't really affect us as much as it might affect somebody else.

No, that's fair. And I think from talking to you guys a few weeks ago, it sounds like your footprint is also overwhelmingly large, well capitalized organizations, you don't have a tremendous amount of SMB in your portfolio, at least as a percentage of revenue. So you'd probably be somewhat insulated from that part of the market as well.

Yes. Most of our customers are multi-metro, multi-region, all around the world. So basically, they have -- we have customers who have -- own no physical infrastructure other than the equipment that sits in our buildings, right? So it's a binary. Either you're in business or you're not in business, right, if you're that kind of a company.

Yes. Fair enough. So I also wanted to talk about kind of the bread and butter of Equinix, which is your interconnection business. So you've materially scaled your virtual cross-connect platform or your Fabric the last number of years. And maybe you could talk about like customer use cases you see that are adopting a virtual versus a physical cross-connect type of product and where -- what those use cases entail. Is it really based on kind of bandwidth demand or capacity needs? And do you see customers adopting both depending on the application? That would be interesting.

So I'll discuss this sort of across multiple verticals because what I think surprises people sometimes is either they think, when we launched this product, it was primarily an enterprise product. And certainly, a lot of our enterprise customers use it. But our service providers customers use it just as much as the enterprises do. So I'll start on the enterprise side because that's a slightly easier one. When cloud came out, most network operators' reaction to that was to say, I will establish an NNI, a network interface with that cloud provider because I've done this thousands of times with other telcos over the years, and almost everybody told us to go away. And 2 years later, they all asked us to come back because they said it's brain damage trying to figure out how to negotiate all of the interactions and the automation necessary to do that part of the business. So a lot of the service providers have essentially said, hey, you're my primary aggregation method for connecting my private network resources to public IT infrastructure. So that's kind of one super big use case. But then the next thing that happened was large enterprise customers of those same service providers came back to them and said, well, hey, I live in an Equinix facility, I buy my network from you, but here's the thing. If I buy a fabric port, I can get access to hundreds of other counter-parties, but not you, the network operator. So that then started driving behavior that says now most of the large -- and Verizon is the big public one. We're also doing this with AT&T, where they sell access into their network out of the Fabric ports that sit in our facilities. So it simplifies sort of all the operations management that, that enterprise has to deal with to say, hey, if I connect to 3 clouds and 3 service network service providers, can I do that on an automated basis? And the reality was everything about that part of the telco business, historically, private networks, private IT infrastructure is highly manual. And so our entire investment thesis around Fabric and Metal would be the other thing, is the idea that says how do we make the private IT infrastructure as agile and easy to provision as public infrastructure. Because public is easy, it's all public IP and cloud. But the reality is there's no deterministic performance and no SLA. And the reality is the service providers make way more money on the other side where there are latency requirements and there are SLA requirements for that particular customer. So being able to do what I call angels dancing on the head of a pin in an automated fashion, when we're already their biggest aggregator for cloud and they've already adopted the automation for that, makes it far easier as an industry to say, hey, I'm going to do this with Equinix because I get instantaneous access to the most counter-parties on a programmatic basis, right? So that's the...

And you can turn up and down capacity on an as-needed basis as well. So it's kind of a cloud-like consumption versus physical cross-connects. And I would imagine, too, there's probably a breakeven point based on bandwidth demand where it does become more economical to compress to the fiber cross-connect.

Well, and what you'll see with what we do is if you're, say, a very large enterprise doing business with one of the large clouds, yes, you will buy a cross-connect because you're pushing so much traffic back and forth between you and them. However, the reality is the long tail of the market from -- and we benefit from that because that could be a customer who doesn't even sit in an Equinix facility. Yet they're sitting in a building at the end of an operator's network, and they're pushing traffic back to us. And so from the operator's perspective, they don't know which cloud they're going to connect to or how much private infrastructure they have. And so the ability to do that on an OpEx basis, turn it up, turn it down is way more attractive than saying, I have to physically provision a cross-connect every single time somebody wants some sort of counter-party to that network PoP. That's manual and makes sense economically at very, very high volumes, but not so much in the middle and small end of the market.

And in terms of your Fabric product, is most of the traffic still occurring within a single campus? Or do you see more and more kind of traveling across your different data center campuses either regionally or globally?

The majority of it is still local. However, the fastest growing components are the long-haul components. And I can describe that from a use case perspective. So what we run into now is, I'll give an example, U.S. network operator doing business in Australia, their primary PoP is Sydney. However, when they do sell a multinational customer who's in multiple metros, they're in Equinix facilities across Australia. So they have one of 2 options. They could do their normal type 2, wait 3 months to get a circuit provision by the local provider. Or they can do it with us on an OpEx basis with no commitment and still convert it over to that circuit when it comes in, but they don't delay service activation for that customer. So that gives a nice sort of flywheel effect to what Fabric enables as far as getting people activated fastest. That is the single largest demand that we get from customers is faster.

Interesting. So you touched on Metal a little bit, but I wanted to explore the Packet acquisition and how that's trending with your bare metal product. And I think early on when you first announced it, some investors said, well, it almost feels like they're competing against other service providers who are in their facilities who can offer similar products. But I think you guys have kind of shown that you're really just kind of expanding an addressable market and solving a pain point that customers really asked you -- were asking you for. So maybe you could talk about the success of the bare metal product so far, how you see that scaling kind of longer term?

I'm happy to start. From a metal perspective, I'd say, as you noted, it's really about giving customers the building blocks of how they want to architect their infrastructure with us. So as the world's digital infrastructure provider, we have done colocation interconnection for almost 25 years. And now with Equinix Metal, and then you combine it with Equinix Fabric, it gives customers the ability to scale and have flexibility to deploy around the world with us without having to shift their own gear to our facilities. So it gives customers the ability if they want to test out an application in a given market without having to send their gear to our facility and spin that up, they can just a few clicks of a button, and use that and metal is deployed in about 23 markets globally.

And so another good thing to think about is, it's almost like, in some sense, a colo proxy, right? So the natural behavior of our enterprise customer base is land-and-expand. So very common starting 2 metros, then they're in 6 metros, then they're in 10, 20, 30, so on and so forth. Well, what's the gating factor to that expansion? The gating factor of that expansion is the CapEx for the equipment that goes in the racks, not us. We're actually a very small portion of the total BOM of any project we've ever seen with an enterprise customer. So the idea here is then to say, hey, now that virtualization is sort of dominant in the market from an IT perspective, the ability to turn up another instance of an application in a remote market where you may have users but you don't have infrastructure now becomes possible on an OpEx basis. And the same thing, if you hit a crossover point where it makes sense to go on, well, where are you going to put that? You're going to put it in our facility because you already have all your traffic and all your counter-party dependencies already wired up because you've depended on us to do that for you. So that's a pretty durable distinction.

And I would have to envision, too, with all the supply chain issues we have around equipment and chip availability that it's also -- your product may be more attractive from a kind of speed to market perspective as well in the current environment.

Yes. We definitely have seen cases where people are having to wait longer. And so they like the idea that says, "Hey, we're a much bigger entity. We have a bunch of better buying power with our partners. And so therefore, we've bought it way ahead because we sort of understand the natural dynamic of our business and how this fits into that sort of existing enterprise base." So being in all 23 markets with ample capacity is definitely working in our favor.

And I think we have time for one more question. So I wanted to ask about potential M&A, and either of you can take this question. I think you've done a lot of geographic-related M&A moving into more and more emerging markets over the years as well. Is kind of what I would call technology-related M&A something that could be on your road map? And I guess, as you think about products that aren't in your portfolio today that may make sense over time, are there any that would be attractive in terms of maybe even moving up the stack a little bit more in terms of what services you offer customers?

Yes, I can start. I'd say from an M&A perspective, overall, we don't comment on any M&A. But when we think about the kind of the things that we look at for an M&A transaction, number one would, as you noted, be expanding our geographic scale and reach. We have been fairly acquisitive as a company. Over the course of our history, we've completed 29 acquisitions. So most recently, we expanded into Chile and Peru with our Entel acquisition. We entered into Africa this year with our MainOne acquisition and then about a little over 2 years ago -- about 1.5 years, we entered into India with our GPX acquisition. So it's a tool in our toolkit to continue to expand our geographic scale and reach. Two would be any time any highly interconnected assets come up, you can imagine we'll take a look, but those, everyone probably knows, [indiscernible] these days. So those are becoming harder to do. And then lastly, on the technology side, it's a -- not looking to step further up the stack. We are the world's digital infrastructure provider, recognizing there's many great companies that operate higher up in the stack. But not to say if there's something that couldn't fit within our digital infrastructure services portfolio over time, but just being mindful we're not looking to move further up the stack. Do you have anything on that one?

Yes. I mean the other way to think about where we've invested that's not physical real estate, so if you think about Fabric and you think about Metal, like you said, they're essentially colo proxies, they're digital infrastructure building blocks, and as it turns out, sort of the software automation necessary to make a highly dynamic environment. Historically, even though we use all the same technology that say a network service provider does that on a wide area basis, the business dynamic that drives the development is very, very, very different. So we don't have to go up stack, right? There's a lot of richness to be unlocked just through whether it's organically or potentially inorganically in the lane that we play in to make the consumption faster, back to my earlier point, that tends to be where we invest our thinking as opposed to trying to compete with a managed -- a full end-to-end managed service provider. That is not what we do.

That's fair enough. Well, that's a great place to end. So Jim and Katie, thank you for joining us today.

Thank you.

Thanks.

Appreciate it.