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

Yes. So no particular order. As I just talked about, Dallas, that is a market that we are incredibly bullish on just the fundamentals there. Just we've got a highly interconnected asset in [indiscernible]. We have several plots of land under control. Power is obviously -- I want to say in abundance, but it's far better than some other markets. And so -- and just the way the Dallas market is shaping up, you're seeing financial services coming to that market. And so it's just a very strong business, which we like. Over to Chicago, the stronger financial services. And then Silicon Valley is an interesting one because it's an expensive market, candidly, for us in terms of build and energy. But we have some customers there that they -- if they want to be there, they really want to be there. And if you think about the spectrum of Silicon Valley start-ups, but some of the new organizations that are popping up, they -- and I was surprised by this when I started in this role, but they are really wanting us to continue to grow in that market. And so just between those three, you've got three very different characteristics. You've got 3 very different governments. How they're thinking about things and different priorities. And then you layer on L.A., Seattle, Colorado, there's a lot going on.

Well, that's something that I want to ask you about how flexible our customers being in terms of their deployment. So are they location agnostic between regions? Or do they have to be in a Silicon Valley if they want to be in Silicon Valley?

It depends on the workload. And so if I think about Silicon Valley, we've got some customers that want a showcase right? And they want that showcase to be in San Jose. So it's close for their customers, their team and potential investors. And so they really want that proximity. And then you go out to other companies. L.A. actually has some characteristics around media and content. There's a strong subsea component there. And so it really depends on the workload. The one thing I would say is that customers are learning. They've got to be -- they need to be a little bit more fungible in their requirements because the capacity of exactly where they want it, when they want it. It's not just given the constraints, that may not be there. And that's going to force them to be a little bit more flexible.

Can you talk more specifically about like those type of workload that would require something that's closer to the market they desire versus something that's further away?

Yes, sure. I mean if you look at Chicago, we have a number of high-frequency trading firms who are there. They absolutely have to be in that downtown area. However, their back office, that can live out in a place like Manuka. And so -- and we work very, very closely with customers and try to help them understand and not everything needs to be in the one place and by bifurcating those workloads, they can ultimately, it can be better for them economically, and it enables us to put the right workload in the right location.

If we switch gears a little bit, Equinix added -- I think it was 9,700 net interconnections this last quarter. I think it was in record for the company. Obviously the tales of AI really taking off. What would -- how would you sort of characterize what's driving that new level of interconnection net adds?

Yes, absolutely. So we've talked about it being a core differentiator and a core component of our business. that is becoming even more relevant as sort of demand take shape in new and different forms. A lot of the demand that we've seen from an interconnect perspective actually relates back to what Guy mentioned, which is it's being driven by not only sort of the traditional customer base but also by more traditional workloads with sort of a layer, maybe an accelerator layer on top of that, driven by the early innings of AI demand. So we think the outlook for interconnect is strong. We're already seeing strength. There typically tends to be a bit of a lag between deployment of space and power in the interconnect associated with those. So we think the outlook is certainly favorable. In customers increasingly need distributed workloads and data. They need interconnection and we have more to offer them than anybody else in the industry.

Yes. Can I just add on to that as well. The other thing that we're seeing is kind of similar to what we saw with cloud. But there is a density forming from a network perspective around, say, new clouds, around, say, LLM and the focus for some of those have gone from where is the compute that I can get access to now how do I distribute that compute to my customers might uses, et cetera. And the fact is that ton of that network is going to happen in and around annex. And so we're kind of seeing this coalescence of density just forming in some of the interconnection markets that we've got a natural advantage in.

Do you think the Neo clouds and the LLM providers, do you think they require sort of a structurally different level of interconnections versus what we saw in the past cloud cycle?

Potentially. Potentially, the agentic side of things is interesting. That's going to -- again, I think people underestimate the network lift and complexity that's required to really deliver agentic at scale. But a lot of that will also happen inside a physical cage or inside a physical data center. It's not like that needs to move around the metro. So I think it's wait and see, but there's no doubt that AI is going to drive a ton of interconnection.

Speaking of AI, obviously, Equinix is very much known for more retail colocation. How do you think about the balance of investment between hyperscale campuses and more of these highly interconnected campuses within the broader capital capital spending?

Yes. So I think in what we've communicated is that we really like our core target market, right, which is colocation focus. No doubt, we have a moat there, what leads to competitive advantage across essentially everything that we do. There are barriers to entry, as I mentioned, both from an operational perspective and from a new supply perspective, and I also mentioned the mid-20% returns that we achieve in our development, right. So we're developing at high yields, but we have a very good understanding of the risk profile in terms of the capital that we're being deploying. So we'll take that risk return profile all day long.

So another thing that you guys have called out, I think, is a high number of sort of neo cloud customers. How do you make sure that you're attracting those customers into your campuses and making sure that they have sort of what they need so you can drive interconnection?

Yes, we're very thoughtful about that. So we have a list of customers that we want to pursue in every metro, and that's one of the things that I can to oversee. Because in every metro, it might be a little bit different how you want to attract to L.A., may be different how you want to track into Dallas. There's going to be a lot of overlap for sure. And so we're very thought about ensuring that we can attract the network component at a minimum of those customers in our facilities. And to be clear, this isn't the training side of things. This is from inferencing forward in the stock. And -- but that is something like we did with cloud like we did with financial services that we're very intentional about converting.

Got it. And I will say we have about 5 minutes left. If anybody has any questions, I'll ask them in a second. But I want to ask you about power. Obviously, you guys need a lot of power in terms of what you guys are going to develop from a facility standpoint in the next couple of years. Where are you seeing the tightness in terms of power availability? Is there a specific geography that you guys think is tighter or looser from a development standpoint?

I'm going to start with my region then kick it over to Ryan. Look, so the power that we've announced, we have largely under control. And so it's not so much worried about that. The other thing I would call out is that a lot of the sites that we're talking about. So say, for example, Dallas, we're looking at a 60-megawatt build there. The threshold that the government have released is 75-megawatts. And so a lot of our builds are still very large, but they're not in this kind of hyper campus piece that I think people are most concerned about. These are also markets that we've been present in for 25 plus years, and so for us, it's really just a continued expansion of our core business. So it's not like there's a ton of AI all of a sudden happening inside the [indiscernible]. It's going to be happening at other locations, but the actual inferencing, the networking components of it are coming through Equinix.

Yes. And the only thing I'd add is that we've talked about having 3 gigawatts of land and power essentially secured, meaning it's they're signed and secured or it's maybe a page flip away from being signed and secured. We're very -- where we've always been very thoughtful about how we do that, how we go about that. And we're also thoughtful about how we communicate that to the market in the sense that most of what you see in our reporting from a pipeline perspective, we have a pretty good degree of certainty around it getting executed.

When you think about like the data centers that you're going to be building today and in the future versus what you had in the past? How do those differ between those sort of different generations?

The main difference is around the power density. So what customers are now showing up for in our next-generation builds. Is they're wanting densities of 40 to 60 kilowatts plus per cabinet. And so that fundamentally changes how we think about designing the site. The other piece would be liquid cooling, it's not going to be all indoor, 5 years from now, but the fact is it's going to be taking up a larger percentage of our sites. And so we're building now with that thinking in it.

And a lot of what's playing out from a demand perspective, AI and otherwise depends on actually some of the stuff, some of the workloads and the networking that exist in the older facilities. So what we -- what we're doing is trying to take advantage of a multiyear period here where we can develop new properties and come out of that period with a full portfolio from the older properties to the newer properties. It's fully set up to serve demand into the longer term.

And how are sort of the utilities responding to sort of your guys' demand? Like are they being more responsive to the loads that you guys are requesting?

We've got great relationships with the utilities. Again, we've been operating with them in partnership for 20, 25 years in many cases. The other thing is that the -- our business model is quite predictable, right? And so we're able to show up in a market and say, "Hey, we're going to need [indiscernible] megawatts demand. We think we're going to need it over this period of time, and you can kind of partner with utilities like that. We're operating on similar time horizons. We're not showing up in the market saying, hey, I need 300 megawatts tomorrow. And I think that's what's really causing the congestion in the system, not so much the Equinix side. And so -- we've got very good relationships across the board with those guys.

Have we got any questions in the audience. I'll probably go one more. As you take on this new role, obviously, the demand environment very strong. You guys have a plan that you've outlined. What risk factors do you see? And what keeps you up at night?

Yes. I'd say probably two things. Supply chain is something that we manage about as best we can, and I feel very good about where we sit there. However, there's a lot of [indiscernible] a lot of the same things. And so that is -- that is something that we manage incredibly closely and generally feel good about it, but there's a lot of complexity that goes with that. The other piece that I wouldn't say keeps me up at night, but from a people perspective, a ton of these things are being built. There's not enough folks. And so we spend a lot of time being very intentional thinking about even from middle school onwards, the communities that we're going into or operating from. How do we educate folks about what a career in data centers look like? How do you progress that through high school into particularly vocational colleges, increase the talent pool that we have because at the moment, it's very constrained. And so that's something we put a lot of energy on.

On community engagement. Well, guys, thank you very much for being here, Guy. And Ryan, I appreciate your time.

Thank you.

Thanks, everyone.