Okay, I think we'll get started. So, last fireside chat of the day, the entire conference. So, thank you, everyone, for waiting it out this long. We're saving the best till last. Then, not me. So, I'm Rob Floyd. I'm in Equity Sales in London. I'd like to welcome Katrina Rymill, Senior Vice President of Corporate Finance and Sustainability, and Philip Konieczny , Senior Vice President of Revenue Operations of Equinix, the largest data center company in the world. Now, I think the vast majority of people know about the company, but there's definitely going to be a minority that don't. So, Katrina, maybe you could just start by just giving us a brief introduction into Equinix, just to level set, and then we can go from there.
Happy to jump in, and thank you for the nice attendance, given we're the last slot of the day. So, great to see many of you in the room today. So, before we jump in, some of what we'll be covering may contain forward-looking statements. If you have any questions, please see our SEC filing. Now, many of you are familiar with Equinix in the room, but just kind of a quick snapshot. Equinix has been operating for over 25 years as the world's leading digital infrastructure company. Our core focus is around data centers. So, today, we have over 260 data centers around the world across 70 metros and continue to expand from there. Now, Equinix is a bit unique in the data center space. If you go back to our founding, we focus on what we call ecosystems.
This is, think of it as companies who find value from colocating, physically being next to each other inside of a data center. The original foundation of the company was really focused on the network piece, being that last mile of network connectivity in these population-dense centers. We focus on retail colocation. So, retail means hundreds of customers per data center, and we build out around networks. The next ecosystem to evolve was the cloud providers coming in, financial services, really creating this environment where there's buyers and sellers of services. From a financial perspective, it's a highly recurring model. About 90% of our revenues is recurring in nature. We just delivered our 85th consecutive quarter in a row of revenue growth, the longest of any S&P 500 company.
If you look at our guidance for the year, we're expecting to grow revenue, as we said on our last Q1 call, 7%-8% from a revenue perspective, more coming down the bottom line, about 8%-11% on our AFFO per share.
Thank you. So, maybe let's start on the demand side of the equation. So, every conference call you listen to, every other word at the moment within the tech community, at least, is AI. Maybe every third word is data center, and every fourth word is GPU. So, it feels like a really hot time to be in the data center space. So, maybe you could start by speaking to the demand, the major drivers of that demand, and sort of how you're thinking about that demand over the next few years and how it develops.
Yeah, maybe I'll just start kind of at the high level that, overall, as you said, we are feeling every other word is around AI or data center, which is great for us. We feel like our relevance with our customers is only continuing to rise because of many of the evolutions that Katrina has talked about. But as enterprises and customers are continuing their own digital transformation journey, they're evolving again how they're thinking about deploying their applications, how they're thinking about the usage of cloud, how they're thinking about the usage of AI. And we really sit at the intersection of that.
When you think about ultimately how that's going to manifest itself in use cases for end users, whether that's businesses, B2B, or with customers in the B2C world, we believe we're kind of that best manifestation of that kind of digital edge where all that traffic needs to ultimately land and intersect. So, we're really seeing kind of the relevance of our presence really rise. And as Katrina said, being in over 70 metros around the world, we're going to be able to play with that global landscape definitely in mind. From a pipeline perspective, I mean, we just talked about the pipeline is, we're continuing to see a very healthy pipeline across the business. We talked about in Q1 how we had the highest Q1 bookings we've had in the history of the company, which obviously is great momentum as we enter this year.
And we're seeing continued momentum on the broader business, kind of broad brush across our broad pipeline. As it relates specifically to AI, I mean, I think we talk about it's still incredibly early days. We're clearly seeing some bifurcation of workloads around the AI front when you think about what's happening on the learning side and the large language models versus what we're seeing on the inference side of things. Obviously, a lot of hype and a lot of money being poured into the space more broadly. But as we think about where we're going to, we think the inference side is where we're going to play more of a role, ultimately in our retail facilities. We've got an xScale offering, which plays more towards the wholesale side of things and can be serving a lot of the AI demand from kind of a large language model perspective.
And we've talked about how we've done a considerable 90 MW of bookings over the last nine months approximately and continue to increase our relevance on the xScale side of things. So, AI is early days as it relates to kind of more of the retail side. And so, I think there's some evolution to go around that. We do see, maybe one last point, general kind of cross currents in the business. I think Charles and Keith talked about this on the last earnings call where we do still see enterprises trying to wrestle with some of the more macroeconomic and environmental elements out there, whether it's the higher inflation, which has obviously come down now, but the higher interest rates and their IT budgets specifically. So, a lot of the AI workloads that they're looking to deploy are oftentimes replacing existing IT applications that they're working through.
And so, we're seeing some of those cross currents where it's not all net gain all the time. And so, we're having to wrestle with some of that with our customers more broadly. But big macro backdrop, we feel really good about the long-term prospects of where we stand.
So, you talked about record backlogs, which I want to come on to, but one point you just mentioned was that you think you'll be more sort of aligned to the inference side. Could you just give some detail as to why that's the case?
Yeah, we think ultimately when you think about how AI is going to be consumed and used and the large language models aren't terribly latency sensitive today. You think about the world today, the initial instance of when ChatGPT was launched was using internet data from a few years ago. And think about a world where that's going to continue to evolve, obviously, but that doesn't have a huge amount of latency sensitivity. When you think about how that's going to ultimately get consumed again by the end user, those will have much more latency sensitive applications, we believe, just as that's evolved from a cloud infrastructure perspective. Those applications that are often drawing upon the usage of the cloud and how that's being consumed does need to live kind of more at the edge.
And then again, when we think about our distributed data centers across 71+ markets and another statistic, we've got 64% of our recurring revenue is by customers who are deployed across all three regions. So, our customers really appreciate that we are distributed across the broader landscape. And I think we think a similar thing will play out on the AI side with inference, although we're incredibly early days on that. But again, when you think about where that's the latency sensitivity is going to come into play.
Makes sense. Okay, so let's just jump back to the record backlogs and the pipeline. I mean, presumably that gives you incredibly strong pricing power. So, could you maybe talk about that pricing power and how sort of new contract renewals work, whether it be an AI application or anything else?
Yeah, maybe I'll start with the pricing side. I mean, first, I would say that we take a very kind of holistic lens at how we look at pricing, especially with our customers, knowing that 90% of our new bookings every quarter come from existing customers. So, we're very thoughtful about how we think about that. But yeah, the overall pricing dynamic that we are living in today is a healthy one. We're seeing really, and from our metrics, what you see how it manifests itself is in our, we call it our MRR per cab or our recurring revenue that we have per cabinet. And there's a couple of nuances in there that it does depend on the density of the deployment. What can impact that? Our interconnection revenue can depend on that. But we were up $119 year-over-year on that metric.
And all three regions, we're seeing more over $2,000 of MRR per cabinet. And so, that's just, and that's been on the rise. And so, again, it speaks to the overall healthy pricing environment that we're dealing with. And I think there's also, when we look at the situation with power more broadly across the globe, we do believe that with AI, as you mentioned, and the consumption of power that that's taking, especially in the large language models, that it's putting overall kind of at a macro level pressure on the consumption of that power and the scarcity value of that. And so, that plays into, I think, the overall pricing environment as well.
Yeah, I would just add on. I do agree. I think some of, actually, the tightest pricing environment that we've seen in years, the last two years, and it does flow through in the MRR per cabinet. For a long time, on the IR side, we'd say flat or firm, and now we're seeing a steady trend up. It's a combination of densification, pricing power, selling more services like interconnection, and just overall a tighter supply with strong demand. So, we've been. I'd say we were hitting that P side of the equation a little bit firmer this year, and it's coming across in those metrics.
When was there a tighter pricing environment? I kind of expect you to say this was the tightest, and if it's not the tightest environment.
No, I think it is one of the tightest. I mean, I would say it usually decouples a bit more by market. So, sometimes you'll have different dynamics where market will look softer and then it will tighten up, but it just feels like an unusual period right now. I also think during COVID, there was a lot of growth on digital infrastructure, and that remains to be the case. And we often get asked about cloud. I think cloud remains a fundamental driver of our business. It may surprise investors to hear. If you look at that top 10 list, eight of the top 10 customers are now cloud providers. It didn't look like that 10 years ago. And we are, you hear cloud, but keep in mind, really Equinix enables access to cloud.
The hybrid cloud, multi-cloud is absolutely our friend and continues to drive our business, and that's been a 10-year trend.
So, I do want to talk about the competition side of things and relationships with Hyperscale because it feels like a sort of frenemy type relationship, perhaps. So, we'll come back into that.
We have a lot of frenemies in the space.
Yeah, I'm sure. But let's stick with power because you guys charge by the kilowatt. GPUs use a lot of power. AI racks use even more. This is all good stuff. Phillip, you touched on power constraints, and there's a great quote from the CEO of Exelon who run utilities in Illinois and Pennsylvania that AI is expected to drive a 900% increase in power demand from data centers just around Chicago, which is four nuclear power plants. So, I guess my question is, at what point does power availability become a constraint, and what are you guys doing about that?
So, let me start with this one because I do feel like one, there's so much buzz around AI, and there's a lot of buzz around power. And I just want to be a little bit cognizant. There are some pretty wild numbers flying around. Like when you look at the actual power consumption from data centers, it's been about 2% in the U.S., and it's actually been pretty steady. So, it's not to say it isn't growing and the whole base is growing, but you have offsets, which is even with the shift over to digital infrastructure, digital equipment becomes more efficient over time. So, the amount that you can compute both from a power perspective and just amount of infrastructure continues to improve over time.
So, while we do see power densities going up, and I'll give you some stats in a minute to give you a sense of that, I don't see it hockey sticking up. So, 25-year history here, we used to build at 2 kVA per cabinet. About 10 years ago, we moved up to 4 kVA. Now we're building about 6-8 kVA on the new build. So, it is trending up. We are watching that. We are densifying. But keep in mind, we can accommodate a very power-dense customer next to a network who tends to use only 2-3 kVA. It's the beauty of the retail model that you're essentially playing a game of Tetris to max out the space and the power consumption. And what we sped up is anything over 20 kVA, which is quite a lot of power, requires special cooling.
And so, that's where you start going from air-cooled. You have to move over to liquid-cooled. We've sped up the investment behind that to enable liquid cooling solutions in over 100 of our data centers.
To put that in context, how many data centers do you have?
260.
How long does it take to upgrade to liquid cooling? And how much does it cost?
So, the cost is in the CapEx. Compared to the grand scheme of things, there's an investment there, but it's not massive. Again, we already are set up for that to enable liquid cooling. And it's really kind of based on customer demand. Not all of your data centers would you initially want liquid cooling. Again, it's a small subset. You're not going to walk into one of our data centers and the whole thing's not going to be running on liquid cooling. It's not today.
Then just finishing, I guess, from the power theme, how about, actually, access to power? I mean, I guess you're not the only ones who, when a new power station goes up or whatever, you're not going to be the only ones who are after that. The hyperscalers, big companies, deep pockets, enormous power requirements. How do you ensure that you get access to that power at the rates that you're used to, or has that landscape slightly changed?
Yeah, maybe I'll start. I mean, listen, obviously, we've got long histories in many of the markets and with the utility providers that we're working with. And if you look at the demand that we need from a power, thinking about from a retail, we call it kind of the retail data center side of things compared to what's needed on the wholesale side when it can be hundreds of megawatts, we need a relatively small fraction of that to run our facilities. And I think we oftentimes are working with the authorities in the different markets. Singapore is a great example where we've been allocated some additional power capacity where that was from a global perspective, one of the markets that had a lot of attention where they had kind of clamped down on the power perspective.
We were talking to them a lot about the power that we bring of helping their overall economy because of these back to the ecosystem story and how we're really helping those communities of interest out. Also, our sustainability story, which Katrina can tell you a lot about as well, really contributed to our ability to secure power from this very constrained market.
And so, I think when you look at kind of the factors of the value that we play in the overall and helping oftentimes local enterprises out and them running their business and the difference that we have and what's required from a power perspective in our model versus in the wholesale model, we feel really well positioned to be able to secure the amount of power that we need, which is smaller relative to the grand scheme of demand that we're seeing across some of these metros.
Are you seeing extra competition coming in from Hyperscalers there?
I wouldn't say it's more extra competition. Obviously, there is that landscape. I mean, there's a lot of capital obviously flowing into the space, whether it's private equity money that's coming in. But again, that's a space where we compete with, we have an xScale product that is our wholesale thing, which is an off-balance sheet joint venture where we bring in partners to help us build out that side of the business. And our focus really for most of our capital is deployed is on this kind of high value-added retail side of things. And we're highly differentiated in that space. So, we really try to make sure that we're talking about these markets in different ways. They have very different pricing dynamics. They have very different return dynamics. To give you a sense, we have 180 assets in our stabilized portfolio.
They have a 26% cash-on-cash yield in that portfolio. We target returns, an IRR and returns of our new builds in that same range, in that 20% to kind of north of 20% range. Very different than the dynamic that you see in this wholesale and Hyperscaler side of things. I just want to make, it's very important to understand there's kind of two different markets that are at play that we're navigating.
So, I guess to that point, the CEO of CoreWeave was interviewed on Bloomberg News this morning, and they were talking about the fact that they are the AI-ready data center. They've raised $12 billion in the last 12 months. Is this just CEO bluster? Do they do something drastically differently to you guys, or are you also AI-ready as a data center company in the same way that they are?
Well, CoreWeave is actually a customer of ours. So, when they locate within us, they're AI-ready. But the majority, I mean, kind of to that point, CoreWeave is an example of an important magnet. That's exactly the type of some of the new companies coming into being that we want to make sure that we're working with. We have a strong partnership with NVIDIA, right? So, they're deploying their DGX pods within us and manage AI as an infrastructure services. So, I do think when we think about our strategy, it is winning some of these new providers. But are we going to supply all of CoreWeave and these other providers infrastructure answers? It's absolutely not. In fact, the vast majority of our customers own and operate their own data centers.
This is where you get into this model of you have a small footprint with Equinix in the downtown locations to do the connectivity, and then you have your own data center further back. So, we don't necessarily, again, we're winning it on the xScale side, but we're not going to win every single element. What you're really seeing too is just the hyperscalers are the ones really moving ahead with deploying AI infrastructure. So, that's working with whether it's Oracle or Microsoft or Amazon, we've seen a pickup in activity around that.
That's with you as well as within their own data centers?
That's right.
Yeah.
That's right.
And like the CoreWeaves of the world, again, GPU as a service types of customers. I mean, we've got several of them as customers of ours because again, they're usually not running their own data centers, right? They're putting it that needs to live somewhere. And that's again, coming back to the rising relevance that we feel like that we have more broadly as the AI ecosystem starts to play out. We don't know exactly how it's all going to play out, but we believe that we're going to be a very important provider to many of the different subsegments in that industry.
Yeah, I do want to be cognizant. I know like in many respects, we are technology, just like we're neutral to what equipment's running inside there, we're also technology agnostic, right? Like we've never, I got questions, you know, are you a cryptocurrency ready data center back in the day? Or are you ready for quantum computing? And are you ready for this or financial services? The answer is we run a multi-tenant data center. So, our job is to be able to accommodate a huge variety of different infrastructure and different connectivity needs. So, part of when we're thinking about, are we AI ready, it's about what is unique about AI, what customers are going to be working with, what power densities, and is there anything different that we need to do that? But it's going to be in this very varied retail environment.
I wanted to touch on sustainability. I guess that's one of your other hats, Katrina. I think Equinix has been at the forefront among tech companies of being very sustainable from an environmental perspective. 96%, I think, of your power is from renewable sources, which is fantastic. But how could you just help us square the circle with massive increases of power demand and how you sort of keep that renewable story on track?
Yeah, the thing I love about working on sustainability at Equinix is it's so closely aligned to how we should be operating as a company, right? True sustainability is just doing more with less from a resource perspective. So, whether it's how we think about driving efficiencies out of our data centers, and it's to our benefit because we get to sell more power, to how we think about renewable energy coverage. And we're really proud of the progress that we've made. And we've set long-term goals. Like we set science-based targets in 2021 for 2030 to reduce 50% of our Scope 1 and 2 emissions. We set a 100% renewable energy target. That's sort of hard stop by 2030. This is our sixth year in a row of having 96% renewable energy coverage. We get that through, to your question, a variety of different sources.
Probably one of the best coverage areas is what's called power purchase agreements. Now, to take a step back, what I would like is the grid to be green, that we can just directly purchase it off the grid and we don't have to come up with these structures. In the meantime, however, as it isn't fully green, what you're doing is thinking about additionality. How are you helping operators add renewable energy sources in the appropriate sectors? And that's these power purchase agreements. We've been really active on the PPAs. It's been a big step up. We have 20 in place. It's going to push up our percentage of coverage over time. These are complex, large, financially liable deals where we have to be very careful how we set these up. It's going to help us hit that 100% renewable energy coverage.
So, the goal is to add renewable energy to the coverage. We're thinking about quality. And then on the local level, work a lot with the utilities, work a lot on public policy. The amount of interest coming in from our customers. We're out here in Europe, the European companies are leading. There's a tremendous amount of interest in sustainability. And the reason is because for those companies like Microsoft to hit their sustainability goals, we have to hit our goals. So, there's a real commitment from our side to help Microsoft and those larger players hit their goals as well.
We talked a lot about demand. What we haven't touched on that much is supply. I imagine building new data centers is very difficult. Not just the access to power, it's the access to land, it's the permitting, it's people that maybe don't want them in their backyard. So, can you talk a bit more about your expansion plans, but also how the supply situation constrains other people? Because I guess one of the strengths of Equinix is you've already got all the best locations signed up.
Well, you know, one of the benefits of being the largest digital infrastructure provider and have 25 years of history is we know how to do this. We've been doing this for a long time. We built or bought 260 of them around the world. Just to give you a high-level perspective, we spend around $3 billion a year expanding our portfolio. That's on organic builds. And we're looking at new markets. And again, as you pointed out, it's the land, it's buying the land, getting the permitting, the power, getting all of the supply chain lined up. But again, being one of the largest in the industries, we're able to kind of line all that up. And we think very much in multi-year horizons, right? This is not, we're not thinking just about 2025.
We're thinking 5, 10 years out as we think about our land strategy, as we think about what kind of power are we going to need from a supply chain perspective. We're ordering a lot of the equipment, which is fungible across our portfolio. We order that well in advance. And so, that's something that we're very well tuned at. And again, we're continuing to grow. We're going to be, by the end of this year, we're opening up several new markets. We should be in 75 markets by the end of this year. And we're continuing to look at how we grow that portfolio. Obviously, some markets are tougher than others.
When you think about some of the markets that are where we've built out quite a bit, but again, we've been thinking very much about land banking in a lot of our tier one metros to make sure we've got that appropriate runway to expand as our customers need to expand with us.
So, we're nearly on the clock here. Perhaps the final question depends how long your answer is, I suppose. But I guess everybody that's involved in, I guess, the AI complex, I mean, NVIDIA is a story to themselves, but HVAC, cooling, electrical infrastructure names, copper, power companies, you know, all those share prices are doing really well. And you guys have talked about demand through the roof, supply constraints, massive pricing power. It seems you're in a really good position. Is there any frustration within the company about what the share price has maybe been doing over the last six months? And is there anything you think the street is missing?
I think the data center business is an incredibly predictable recurring revenue model. Now, to be fair, I mean, our revenue literally goes up like a ruler, right? So, yes, you probably could predict what we'll be trading in the future just based on forward projections pretty easily. We're thinking we're in a very strong, very defensible business. I'm not sure. The more you get to know the space, the more you recognize the barriers of entry into data center. And there's a lot of complexity to that that's interesting. Couple that with just an overall push of digital infrastructure. I think of cloud having driven our business for 10+ years and will continue to do so. I think AI is a new interesting technology trend, but that's another 10-year cycle.
It'll take time to embed AI into how enterprises actually want to consume and get economic benefit out of that. That's going to be a very interesting journey that we will help enable.
Katrina, Phillip, thank you for your time.
Thank you very much.
Great. Thank you.
Thanks.