Good morning, everyone. My name is Brendan Lynch. I cover REITs here at Barclays. I'm very pleased to be joined by Barry Hytinen, CFO of Iron Mountain. Barry, thank you for doing this with us.
Thank you, Brendan. It's great to be here.
Given this is a tech conference, maybe we'll focus more on the tech angles of your business. Just to start off, for people who aren't as familiar, why don't you give us a brief overview of your positioning within the data center space?
OK, so data center is one of our key growth engines of the company. It represents, round numbers, about 10% of revenues, and that's been ramping appreciably. In fact, if you go back five or so years ago, it would have been low single digits, and our data center business today, based on the land and power we have, can build out over time to about 1.25 gigawatts. Interestingly, we're only operating today about 365 megawatts, of which we're almost 100% leased on, and then over the next, you can see it in my supplemental, over the next couple of quarters to a couple of years, we will open up sites that we are under construction on. It's about 210 megawatts that we're under construction on, and importantly, we're in excess of 95% pre-leased on all of that.
You should anticipate, by the way, that we will continue to acquire more land and power to continue to build out our data center platform. In fact, just recently, in the last few weeks, we announced that we had purchased a couple of parcels in Virginia, one in Northern Virginia, which is adjacent to our existing campus, and one in Richmond. Those two sites combined increased our total capacity by about 350 megawatts over time. That's in the 1.25 I mentioned, but it gives you a sense we're continuing to look for more land to build the portfolio. We've been winning with driving our growth through strong relationships with key hyperscalers.
So if you look at the company seven, eight, nine years ago, we were generally a retail colo kind of data center operator, relatively small, and in just a few markets, particularly Phoenix, a little bit in Northern Virginia, et cetera. But over the years, we've been building out our sites, and our early wins with hyperscalers several years ago in small deployments kind of continued to grow such that now we're routinely signing full building leases with hyperscalers on a pre-lease basis. And you think about the 210 megawatts I mentioned that's under construction, nearly all of that, as I said, is pre-leased. That would all be with hyperscalers. And so we are, from a geographic standpoint, a major global data center operator.
We've got significant campuses in Northern Virginia, in Phoenix, in Amsterdam, London, Frankfurt, and we have several operating sites in Asia, Singapore, a couple of two or three in India, and we've got a lot of power and land coming online in India over time. We also have a small operating site that we're developing out in Spain. So we feel very good about the data center industry, Brendan, because the economics have been improving in that space for, I think, all operators, and we feel well positioned with where we have capacity to continue to build out.
When you look at the opportunity to purchase land and to secure power, or the challenges, I should say, of doing so, how are you navigating that process? Where do you see the challenges? Where do you see the opportunities?
Carefully, I would say. Power is a bottleneck for the broader industry. We have been fortunate that I think our teams have been both strategic and probably partly lucky about where we have had land and the relationships we've developed on the local level with local municipalities, together with the local power companies, because we haven't had any, if you will, hiccups in power delivery like some in the industry have experienced and in some markets. In terms of going forward, we are certainly cognizant of the fact that there is a sort of supply-demand imbalance as it relates to how much data center capacity is required by the industry versus how much power is available, so I'll take the two examples of the recent land purchases we made.
In both of those cases, we were effectively under contract but without a condition to close for quite a while as we made sure that we were very comfortable with the ability to have the power companies deliver the power on time for when we'd be pre-leasing into it and when our construction would be. Because candidly, if you look at power today, of course, it varies by market, but it can take anywhere between building out a substation might take six to 12 months, but getting the long lead time transmission lines and various other elements that are required for power, if that's not already present, you're talking anywhere between a year to, in some markets, three or four or five years. So it is a very important portion of the build-out and site selection of land, absolutely.
In terms of the two parcels that you've acquired recently, do you have substations on those? Are you building those out yourselves? How should we think about that?
Yeah. So in Northern Virginia, you've been covering this for a while, Brendan, so you know and you've even seen, we have a substation that we have put up on our existing campus, and that covers that entire campus. We're going to put another substation on with the new adjacent parcel where we are opening in Manassas, and that will feed the entirety of that new parcel as well as give us some incremental lift on the total campus. So that's going to make that campus even that much more powered for us. And in Richmond, it will also have its own substation, and we've worked with the power companies to work through all the planning related to that.
So, for example, we'll even go under deposit with them for various long lead time items that they have in their supply chain to ensure that everything's married up so that that's all factored into our underwriting as we pre-lease.
You mentioned that the business has grown substantially over the past couple of years. I don't think that comes as a surprise to anyone who's been looking at the data center space. How do you assess the risk of a potential slowdown at some point? Do you see that as a possibility?
I mean, I would never say never, but I will tell you, Brendan, and you and many investors know this, there is a tremendous amount of demand out there from hyperscalers for both data centers today as well as over the next few years, and the relationships we have, I would say, are moving from strength to strength. One of the, if you will, secrets of our success is that we have gotten good footholds with key hyperscaler clients. We've delivered on time, on budget, consistently getting them the power.
There are some data center operators, smaller, that haven't been able to do that as effectively over time, and I think that has created a situation where having key relationships from a standpoint of partnership of building out sites for hyperscalers is that much more important to them in terms of track record and consistency of ability to do the work, because it is pretty complicated to build out a data center. I mean, it's not just the construction of a shell. It's also making sure you have the long lead time for the MEP that's going to go into the site and having those relationships with those vendors. And those are long lead time things like gensets, chillers, what have you.
It's also about making sure you have relationships with general contractors who actually know how to do this and all the subs that go along with it, the points around power that you mentioned, and also negotiating and working through the permitting process. There's a lot that goes into the data center. So I think we've got the credibility and the track record at this point that hyperscalers can count on. And when you look at how much power draw the industry expects, I mean, there are estimates of data center industry growing 25% CAGR for the next 10 years. And candidly, it just doesn't seem to pencil that there's enough power to go around to support that level of growth.
I think it's going to continue to be a good environment as a developer and operator going forward for anyone who has the ability to do all of those critical steps to building this infrastructure.
I would certainly agree. Based on all the channel checks we've ever done, consistent execution is one of the key differentiators between the winners and losers.
I completely agree, and I think it's an underappreciated portion of the story, and it's a credit to our team and many other data center operators that they've been able to do all of this on time and on budget. It is a capability and competency that is critically important to the data center, to the hyperscalers. I mean, you think about it, this is infrastructure that is tantamount to their business proposition. So having it on time or even early is a huge positive force to the relationship building.
There's been a lot of talk about the AI training and the building out of those models over the past couple of years. We may soon be entering the phase of inferencing or ramping in the phase of inferencing. How do you see that playing out within your business?
So, I would say if you look at what we're operating, because that's stuff that we went under contract with either as part of our colo business or were deals that we signed one to two plus years ago in terms of the build-out, because just as a reminder, we're not building really to spec on anything at this point. Everything that we've got under construction is built to contract, and it's been that way for us for the last one to two years. And I would say that that book of business is generally more pre-AI kind of time frame.
Some of what we have under pre-lease that we're building out would be specific to AI applications, but we are not one of those players that's working with the hyperscalers on doing like a 500 megawatt campus off on some very faraway place where they don't require concerns around latency and things of that nature, where they're maybe doing some of those really big training modules. Frankly, I see a lot of that being done by the hyperscalers themselves on their own balance sheet. Where we're supplementing is in closest specific markets where they have needs, where we have the ability to support those. And clearly, I think both are going to continue to be big growth engines, Brendan, both elements of the needs.
You mentioned some of the international markets that you have exposure to. The AI training, most of that build-out has been here domestically in the U.S. Do you see that shifting and those becoming more markets of AI-oriented growth?
Yes, definitely we see indicators of that happening, particularly in Asia, but also in parts of Europe, so I would say there's very substantial growth in data center for a variety of reasons globally, yes.
You mentioned earlier that the returns on data center build-outs are improving. Can you just kind of walk us through the past couple of years and what yields you're generating now?
Sure. So I think if you go back three, four, five years ago, most operators in the data center industry were probably writing cash-on-cash unlevered returns in the 7%-8% kind of a range for hyperscalers specifically. There were some deals that I would guess, not deals that we did, but there were some deals we saw in the industry that probably were even sub-7% at that time frame. We never dipped that low, but over the last few years, that's been rising nicely. Now, part of it should because construction costs have been up as well as, let's say, carried interest.
However, it's been expanding at a nice rate, I think largely because of the supply-demand imbalance that we've talked about, which is likely to be pretty sticky for some time, together with our ability to deliver on time and that credibility factor such that I'd say most folks in the industry are now for hyperscaler returns would probably be north of 10, so 11, 12, 13. You sometimes even hear about some operators doing a little bit higher than that. And I think that's also another interesting leg of the story because if you think about our business, most of what we've got operating from a hyperscale standpoint, we're at those lower return kind of underwriting. And as we move forward, we'll be opening up sites that will be at penciled returns that are much nicer.
I think it's also you mentioned this, but it's worth repeating that you're doing nearly all of the development on a build-to-suit basis, so not taking any development risk.
Yeah, that's exactly right. When you think about our model, it has changed appreciably because years ago when we were a colo operator, we were basically setting up a building, and then it would take multiple years to stabilize because you'd be kind of bringing in tenants and a lot of tenants, frankly, on a per-building basis. But at this point, for the last couple of years, the vast majority of our leasing has been hyperscale, and most of the hyperscale deals that we see are 10-15-plus years in duration with embedded escalators and at the returns that I was describing and ramping. So it's been a very good portion of the business has been moving in that direction, and that enables us to be very confident in terms of how we see the growth playing out over time as well as the margin.
Let's chat a little bit about the Global RIM business, which for those of you who aren't familiar, is mostly paper storage, but there's a lot of digital components that you've brought to that business as well. Maybe you can walk us through some of the innovation there.
Sure. And just to give a little context before I go there, the company this year is north of $6 billion of revenue as we've discussed in our guidance, and if you think about it, we have three growth engines that I alluded to earlier, so we just talked about data center. That's pro forma run rate about $600 million of revenue. Digital solutions, which you were just asking about, is about pro forma exiting this year around $500 million run rate of revenue. And then our Asset Lifecycle Management business would be similarly sized. All three of those businesses have been growing and I expect to continue to grow at a high rate. Think like 20%+ CAGRs and likely to continue to grow at or beyond that level. Some of those businesses are growing a lot more than that.
And then you've got our record business, which is where the company got started. We have 248,000 client relationships in our company. Most of them have been measured in many years, if not decades, and we have approximately 95% of the Fortune 1000 that operate with us. And what the team has been doing here for about seven, eight years is they were planting maybe nine years in some cases, planting seeds for other areas of business that could grow and be adjacent to our existing offerings that we could then cross-sell. And so that first and foremost was Digital Solutions, which many times starts with scanning, but then now has moved into our proprietary SaaS model of DXP, Digital Experience Platform. But they were also investing in Asset Lifecycle Management as well as data center.
Over years now of driving those businesses and seeing an additional cross-sell, they now represent 25% of the business, and they're growing at that fast clip I mentioned, so it gives a nice almost media tailwind to the organic growth of the company in light of those three businesses. Now, in Digital Solutions, which is in that Global RIM business today, as I mentioned a moment ago, many times it might start with scanning, but it really is all about helping clients be much more effective in both managing their workflow of documents and various other things, whether that be printed or online, and being much more efficient with it. So we're saving clients a lot of money, and frankly, Digital is growing, and I think it's kind of an open door for us in terms of the opportunity set there.
Importantly, more and more of that business is recurring, like in a SaaS type of model, whereby we are helping clients save money utilizing our platform, our DXP platform, to continuously improve workflow, and that enables them to do things that they previously were never able to do, so for example, if they were storing records with us in a box, they had no access to the ability to do anything with it, analyze it. Now with improved scanning operations and we do meta-tagging and we can use AI and machine learning, we can both improve their efficiency because there's still a lot of paper in all of our clients' business models, together with give them the ability to analyze parts of their inventory that they didn't have access to before.
For example, one of our clients is doing a large implementation with us in which we're helping them analyze medical files to check the, if you will, efficacy of treatments and regimens over time. We have another client on the financial services side where we are helping them dramatically streamline their mortgage processing and be able to both improve the speed to decision, but also on the back end, simplify all of the back end, largely labor-intensive portions of mortgage processing. I think Digital Solutions is a business for us that we're in the long-term frame of thinking about that's a $1 billion-plus business for us. Today it's $500 million and it's been growing at like 25% CAGR for quite a few years.
And you mentioned some of the work orders that you have, but the size of some of these is absolutely staggering. Hundreds of millions of documents, if I recall.
That's correct, and we've talked about a couple of our more recent larger wins earlier this year, which are measured in tens of millions of dollars over a five-year period, which is a pretty meaningful contract to our size of company, and they're coming in at generally pretty good margins, and I think there's more opportunity for the margins over time in digital to grow. That's partly because of that SaaS model I mentioned, and it becomes more predictable also because we've got a portion of the digital business which is recurring now going forward, whereas if you go back several years ago, much of it was kind of project-based, whereby it would be like we do something on a small basis with a client and then it'd be over and we'd have to win another, if you will, scanning operation.
We've really moved up the value chain with respect to what we're doing there, and that's likely to continue going forward for, I think, a long period of time.
Just while we're on Global RIM, obviously pricing is an element that gets a lot of attention. You've been pushing price fairly assertively, I'd say, in the past couple of years after maybe probably underpricing for some period of time. How should we think about that opportunity going forward?
Yeah, so we think about our revenue management strategy as let's make sure we get paid for the value we're delivering. And so it's important with any client relationship that we want to be fair and make sure that we're delivering plenty of value for what we charge for the services. I think one of the things that has enabled us to have relatively more pricing power, if you will, than what maybe some folks expected some years ago was because, A, historically we probably just had underpriced for value. If you look at the company going back more than, say, seven or eight years ago, there wasn't much in the way of revenue management here. Yet we were continuing to build out more solutions and service offerings that make our service that much stickier to clients.
Also, Brendan, as we've discussed before, this is a very small cost line in most of our clients' relative P&L. I mean, you think about it, no client for us is even 1% of revenue. So there's a really long tail of client relationships here. And the reason that many clients have chosen to standardize with us is because we can meet their needs in ways that a lot of competitors can't. Meaning, an example would be we operate in 60 countries. So if you're a multinational operating in multiple markets, we're a natural fit for them in their records. And our retention rates are very high. Think like 98.5%. I mean, once we get a client, especially on the medium to larger side, it gets into a situation where it's just kind of like the boxes keep coming in.
Importantly, we have really good data with respect to being able to see the relative impact of pricing because when you think about it, all of our clients are sending us boxes, let's say, weekly, and so we have the ability to look at what is inbound volume coming in 30 days after, 6 months after, 9 months after a price increase versus what was it 30 days before, 6 months before, 9 months before, and just generally speaking, I think that data bears out that we're delivering a tremendous amount of value to clients, and our expectation is for revenue management to continue to be in that mid to upper single-digit vicinity for the foreseeable future, where it's been running at that level now for a few years.
I think as long as we continue to bring out new services and products that make it that much more valuable, like Smart Sort or our Digital On Demand, the Digital Solutions I was mentioning, the DXP platform, that just makes it that much stickier.
Makes sense. Let's chat a little bit about the ALM business. You just entered this a few years ago. Maybe talk about some of your successes with this new platform, some of the challenges that you've faced.
Sure. Asset Lifecycle Management for us kind of has two distinct pieces of business. The first part that I'll talk about is our data center decommissioning. This cross-sells really well with our data leasing business, by the way, because the vast majority of the data center decommissioning that we do is with the major hyperscale tenants, some of whom are tenants on our existing sites. There's not 100% overlap there, so we've been working on the relationship building to improve both businesses there in terms of cross-sell. Data center decommissioning tends to be what I refer to as a revenue share model, whereby what the clients are having us do, if you think about the major hyperscalers, they'll put up a data center or they will work with a third-party data center, install servers into it, and they generally refresh the gear inside the site about every five years.
That's generally on average the rate. It seems to be coming down a little bit in some modest respect due to AI. And that gear that they're retrofitting, it isn't obsolete. They're changing it out because they want better power draw or they want more compute or they want to move to GPU, that sort of thing. So there's inherent latent value in that gear. So they ship it to us. We wipe it. In some cases, if it's been written off, we wipe it, then we destroy it, or in some cases, they let us resell it. And then we are physically decommissioning the servers, CPUs, DRAM, drives, what have you. And then we sell all of that gear up. And we generally get, depending upon the contract, let's call it 20 points off of whatever we're selling.
Now, that business was really challenging over the past, call it, couple of years because component pricing fell dramatically. In fact, we were underwater on profitability in that business at some points last year. But pricing has started to move up. We've also been seeing nice increases in volumes as we've gotten more relationships with hyperscalers and that cross-sell has come on. And as mentioned in the last couple of quarters, we've been doing a lot better in those businesses and they're going from a negative to a positive from a standpoint of EBITDA and growth.
And then the other part, and I should say, that business is interesting because when you think about it, what we're decommissioning in those hyperscalers, that book of business has been growing dramatically over the last five or 10 years as data center operators have opened many more data centers and hyperscalers have been proliferating CapEx into sites. And so the book of business that we will decommission, say, next year is much larger than the book of business that's available to decommission this year because of the inherent growth. And the stuff that we'll decommission this year was put in service five years ago as either new sites or retrofitted gear from older sites five years before. And so there's a very strong amount of pipeline that we can see going forward, and we aim to be the market leader in that space.
I think the ability to actually do the physical decommissioning, consistency of privacy and trust of security, and also the ability to sell it downstream is critically important to the hyperscale clients because you're talking about a lot of gear. On the enterprise side, which is the other half of our Asset Lifecycle Management , that cross-sells quite well off of our existing records base because that's working with medium to larger size companies in terms of decommissioning their IT gear. So that could be laptops, iPads, phones, screens, printers. And there's a secular tailwind there as CIOs and CISOs and even CEOs are more and more concerned about both ensuring that anything that's been written to is wiped appropriately and also sustainably dealt with in terms of a circular economy or recycled. And that's a business for us that we think has a very large runway.
Bill and the team got started in that business back in 2017, and by 2021, it had grown a lot, but it went from zero to, say, $50 million of revenue. Today, our total ALM business is now pro forma about $500 million. We have been acquiring businesses in the enterprise space specifically. Early this year, we bought a business that was a little over $100 million of revenue, 25% EBITDA margins. We bought that at a 7.5 times multiple, and we've synergized it. It's gone extraordinarily well. What we're doing, and we bought two other businesses here recently. The first one I mentioned is a U.S. operator, and then we recently bought an operator that has presence in Europe and the U.S. and a small operating site in Asia. And we also bought a business that has five processing sites in Australia.
The thing to think about this, Brendan, is historically we were doing kind of the front end, but we didn't have the processing capability to do all of the work that goes on after taking that gear in from the client. What we're generally doing is a build versus buy decision on these acquisitions, whereby we can go buy processing capability and then thereby capture the full value chain of profitability from both getting the gear to processing it completely. That part of our strategy is working really, really well because it creates a lot of synergies. Both we get incremental utilization of the facilities we're picking up as we start flowing the Iron Mountain historic volume through them, but together it gives us more heft and the ability to serve clients that much better.
I think that is a business that has a very large TAM, think like north of $20 billion. It's a very fragmented industry. We are probably already by far the largest player in Asset Lifecycle Management . We aim to be the market leader. I think it looks a lot like the early days of our records business in that it's a large market, it's growing, and we, I think, are in a very good position to become the market leader there.
Great. It looks like we're just about out of time. Thank you very much, Barry.
Always great to be here, Brendan. Thank you.
Thank you.