Okay, let's go ahead and get started. Good morning, everyone, and welcome. I am George Tong. I cover business and information services here at Goldman Sachs. I am very pleased to be joined by Barry Hytinen, CFO of Iron Mountain. Barry, thank you for joining us here at the 2026 Communacopia + Technology Conference.
Thank you, George. It's great to be here.
Okay, Barry, let's start at a high level. Iron Mountain has evolved over the years from a legacy traditional physical storage company into a broader platform that encompasses data centers, asset lifecycle management, digital solutions. Can you talk about how these businesses reinforce one another and where you are seeing the most cross-selling opportunity today?
Yeah. Thanks for that, George. We are an interesting company for several reasons, and one of them is our client base. We have 245,000 business-to-business clients, and we operate with 95% or more of the Fortune 1000 as clients. Most of those have standardized with us in records management decades ago. Over the years, Bill, our CEO, and the team, they started looking around for where they could put investment dollars behind things that would be natural cross-sell opportunities of that large client base to continue to expand our share of wallet with those clients. If you look at our asset lifecycle management business, which is really two distinct businesses in there, we kind of even run them separately.
There's the hyperscale data center decommissioning business, then there's the enterprise business, which is corporate clients where they would be having a continuous flow of laptops that need to be refreshed or go obsolete, or printers, screens, what have you. That client base cross-sells very well off the core records client base because the vast majority of our largest clients on the record side are big corporates, Fortune 1000. Similarly, those clients have a very distinct need that they're currently having to fulfill through many different small vendor relationships on the IT asset disposition side. We're building out a global offering that they can't get from anybody else.
When we talk to clients, we know that the reasons that they choose a vendor in that area are related to chain of custody, consistency of process, being able to be. Trust is a huge point, privacy, ability to securely wipe data off of gear. When you think about some of the reasons why they standardize with us on records, it's about chain of custody. It's about trust. It's about consistency of process and ability to serve on a continuous basis. So we think there's huge cross-sell opportunity there, and it tends to be a land and expand model. So we might get a given flow on the ALM side with a corporate client, then expand over time into more and more regions or business lines, what have you, in the various customer. So that's a huge reinforcing one.
In our digital solutions business, which is another fast-growing part of the company, it cross-sells off of our core very effectively as well because many times those projects start with a digitization concept. That digitization is something that traditionally we might be digitizing something for them that we're actually storing. Then we can work that into our DXP platform, which is a huge grower for us right now, George . The digital business is becoming more recurring revenue and more Software as a Service with each passing year. In the most recent quarter, the team grew 20% there. We have some very distinct, interesting government wins that have happened in that business as well, again, reinforcing for existing clients.
In our data center business, where we've been growing most significantly over the last few years is on hyperscale data center clients, where we are a core partner to call it a handful of the largest hyperscalers out there. That reinforces and cross-sells very well with the hyperscale ALM decommissioning business. It's really the same client base, a lot of overlap there. We think over time as being the. We can become the only partner to hyperscale data centers operators that we develop, build, operate, and can decommission for a hyperscale client, whatever they need, wherever they need it around the world. So they may not all look like they go together, but they do all very effectively go together.
Mm-hmm. Your data center, ALM, and digital solutions business, all those combined now represent about 35% of revenues. Most recently grew more than 50% year-over-year. What would you say is the sustainable long-term organic growth potential for this growth portfolio and for Iron Mountain overall?
Yeah. For the growth portfolio, we've said for quite a few years that we expect that growth portfolio to grow north of 20% combined, and we continue to see that happening. If I break that down into the parts, when you look at our data center business, our team has built a really nice data center business that's still got a ton of runway for growth. This year, we'll do a little over $1 billion of revenue in data center, and the team has already signed contracts for data centers that we need to build and energize, which would grow the business another, call it 40%. Just with what we've already sold.
Then we've got a very significant amount of additional land that we'll energize over the next few years, George, to continue to grow the data center business. So there's a lot of growth in our data center. In the ALM business, which is a business that in 2021, I think we did $30 million of revenue. This year we projected we do approximately $1 billion.
That business has been growing rapidly, and the TAM, the total addressable market for ALM is very large. It's $35 billion estimated. Now, of that, 75% is the corporate clients that I mentioned earlier. That's an area where we're roughly $600 million of revenue today, this year. There's a huge runway there for additional cross-selling and growth, and we think that business will ultimately measure in billions of dollars of revenue.
Mm-hmm. Mm-hmm.
On the hyperscale side, specifically hyperscale data center decommissioning, the TAM last year was estimated to be $3.5 billion. Of that, 35% is for hyperscale data center decommissioning. If you look at the embedded platform of data centers that are out there that need to be continuously retrofit with their gear changed out over the next few years, the estimate is for that TAM to double. That $3.5 billion will go to $7 billion.
That is a tremendous amount of incremental opportunity for us to get market share and just growth just with the market. If you look at our digital business, George, it has been growing high teens, for most recent quarter, 20%. We have, I think, really tapped into and are benefiting from how important data is
in the broader economy, especially with AI being an unlock to that data, and being able to help clients monetize and become more efficient and more analytical as it relates to all the various content that we can help them manage through our digital solutions team.
Mm-hmm. Let's dive further into data centers. You have leased 110 MW year to date in your data center business. And you indicated that full year, the leasing should well exceed the initial target of 100 million plus. How would you characterize the pipeline today, and what are customers prioritizing in terms of current leasing discussions?
Okay. We initially projected that we would do 100 MW or more, and as you know, through July, we are already beyond that target. And we did say we expect to be meaningfully beyond it.
Yeah.
Why do we see that happening? Well, couple of things. One, we have got a lot of megawatts that are energizing over the next 18 to 24 months that is not yet leased. So, through July, post July, we had 325 MW that will energize over the next 18 to 24 months that we can lease.
Those are almost exclusively in Tier 1 markets, think like Virginia and parts of Europe and India, where we leased a lot of megawatts in July. 51-MW lease in Mumbai. We have very good assets to lease in locations that are very important to the major hyperscalers who have been historically our key clients. And we have got a track record of producing and being able to build on time, on budget for those clients, and then service them very well. And when we look at the actual pipeline, it is very strong, as we mentioned on the last call. We have got a deeper pipeline than we have historically had, and we have multiple parties interested in each one of those buildings that I was just, or would-be buildings that I was just alluding to.
So I would say, George, we feel very good about our ability to lease over the next few periods. I would note that hyperscale leasing, as we have said before, it can be kind of lumpy in nature because it is a full building, for example.
And in some cases, we are talking to clients about a portion of a campus, and then we have other clients who are looking at an entirety of a campus. So we have got a lot of pipeline there at very good returns, and we expect to lease a lot of megawatts over the next few quarters.
Mm-hmm. On the topic of megawatts that are going to be energized, you have got 325 coming over the next 24 months, 18 to 24 months. How much of that capacity is already leased, or in advanced discussions, and what would you say is the typical timeline from energization to actual revenue commencement?
Okay. All of that 325 you were just referencing is not yet leased. None of it is leased. All of that is incremental to our P&L, going forward. Hence my point that we're going to lease a lot of megawatts over the next year or two. The thing to know about conversion to revenue is, generally speaking, we're not a speculative builder, so we're going to commence the construction once we've got a pre-lease.
for those assets. The average pre-lease with a large hyperscale tenant is running 10, 15 years of duration, with good cash on cash on levered returns. We generally seeing our clients lease 12 to 18 months before delivery. The nice thing about that, George, is that we can, generally speaking, build a data center in 9 to 10 months in most of the geographies.
We have long lead time equipment already on order to be able to do that, such that once we get the pre-lease, we will generally be able to construct the asset in advance of the energization and marry that up right so that we're delivering it when the energy is there, so you don't have any sort of situation where you're carrying an asset that's not performing. The conversion to revenue income is quite rapid.
Mm-hmm. Got it. If you look at recent leases in key markets like London, Amsterdam, Mumbai. A lot of it was driven by AI inference demand. What does an inference-oriented deployment require from a data center operator?
Yeah. So, in our situation, in almost all of our leases, they are either for core cloud or inference, as you mentioned. More and more of the pipeline is becoming AI-oriented inference. So what the general difference is they want more density, and they want, traditionally, the availability to do more liquid cooling.
That does increase the price per megawatt to build, but that also increases the relative price that we are getting in terms of for the lease. It is a format that we have built repeatedly and continue to see growth of significance.
Yes. Power is still today one of the main principal constraints of data center development. How confident are you in the power delivery schedule that is supporting this 325 energization schedule, and where do you see the greatest execution risk?
Yeah. We feel very good about our energization schedule and our teams deal with the various utilities that we are working with on a regular basis, like every week. Frankly, if you look at just what we have energizing over the next 18, 24- even, say, 18, 24 months or in the next couple of years beyond that, it is with utilities that we really know very well.
Because we've been operating, for example, in Virginia with Dominion for years and years now, and they are the key utility for both the Richmond and the Manassas market.
We've previously been under deposit with them for certain long lead time elements that they need to build out the grid to support our future data center sites. We feel very good about how that's progressing. We are in communication with them as it relates to their lead times and their timetables in terms of the scheduling. Similarly, we have relationships in Mumbai, for example, where we're building there as well, and in places like Madrid. We're not having to spread ourselves too thin, George, because when you look at where we have megawatts to sell and energize over the next few quarters, it's in very specific markets where we know the utilities well. I feel highly confident in our ability to energize. In terms of other elements, look, you said it a moment ago that power is the biggest governor.
I would agree with that as an industry. There's also a lot of long lead time equipment that one needs when building out a data center. Thinking about generators and backup generators, you could be talking two, three year or beyond lead times. Cooling technologies and things of that nature. Luckily, as being a major operator and one that's kind of looked out, we've got all of that on schedule, and we are in very routine contact with our key suppliers, and as a business that's growing quite rapidly, we've developed, I'd say, very good relationships with those key vendors for the long lead time stuff.
Data center renewal spreads reached 12% on a cash basis in the second quarter. How sustainable would you say the current pricing environment is? Particularly as you look at customers that are pursuing these larger AI-related deployments and given all the significant power developments, how do you see the overall return and pricing environment evolving?
Okay. I am going to break that apart because in our business, the renewal spreads is really as it relates to our enterprise colo book, which is ballpark a third of what we are operating these days. The mark to market and renewal spreads on that part of the business have been very strong for quite a few years now. Think double digit to even 20% compounded for the last three or so years.
That is really just a factor of what is going on in the given markets where we have space, George, because if you look at our relative churn, it is quite low. It has been running, I think, last year we were at sub 3% and kind of like 2% or so through the first nine months, 2.5%. Quite low, and so that speaks to the fact that it is not like we are pricing beyond what is in market. The alternative for our colo clients, which renew about every year on average, is what else is out there.
Frankly, that is the market. As it relates to renewals on our larger format leases, the fact is we are still a very young operator. We did not get our first large single-tenant lease on the hyperscale side until about 2020, 2021, and that lease did not even commence until 2023.
It was a decade of duration. If you look at our lease expiration table, what you would find is we have a tremendous amount of megawatts that do not renew for in excess of five, six years. In fact, our weighted average lease expiration across the entire portfolio is north of 10 years, and that includes the colo book, which renews every year with that low churn. I think in the future, to project out some, all of those client contracts that I was just referring to on the hyperscale side have renewal options for them. My guess is, in light of the fact that energy continues to constrain and will likely continue to be a constraint for quite a long time, it is going to be very advantageous for those clients to renew in the future. But you are talking about five to 10 years from now
Right
or longer.
Right. Okay, let us switch gears and talk a little bit about the ALM business. In the second quarter, the asset lifecycle management business grew nearly 90% year-over-year. You mentioned revenue scaling to basically $1 billion for the year. How much of the recent growth acceleration would you say is being driven by customer adoption compared to, say, memory prices or project timing for ALM?
Okay. Again, breaking the business into the two parts, the way we manage it, you have our enterprise or corporate side, and then you have the ALM data center, hyperscale data center decommissioning. In that side, in the hyperscale decommissioning side, is where you have a lot more exposure to memory.
In any given period, memory might be 50% or more of what we're decommissioning in terms of the value of the gear. To be sure, memory pricing has been up year-on-year, and depending upon the grade of memory, because there's many different, if you will, SKUs of memory based on the manufacturer of the memory or the relative size, speed, et cetera, pricing has been rising. Some prices have been, as I mentioned on the last call, depending upon SKU to SKU, you've got some that have been up as much as 30% in recent periods. You've got others that have been down some. What really also needs to be in the calculus is the relative mix. As I mentioned on the last call, in the second quarter, I think some investors had expected pricing to be a really big benefit to us.
It was on some SKUs, but in total, we had relatively more DDR3 in the second quarter than we did in the first quarter, when we had more DDR4. DDR4 is at a relatively much higher price than DDR3 because it's a newer generation. So there is an element of mix involved here. Look, I projected out going forward for this year that memory pricing would be basically pretty stable for the remainder of the year from where we exited the second quarter, and it's been probably stable to even slightly up since that period of time. Again, the mix is a big factor. I will note that on the hyperscale data center decommissioning, that's a generally a lower margin business for us versus the enterprise side.
On the hyperscale side, you're talking a revenue share model where we might get 20% of whatever we sell the gear for and the rest goes back to the client. That's meant to cover our costs of decommissioning, fulfillment, shipping, et cetera. So the margin structure on that business is more like low double digit to high teens, that kind of thing.
But on the enterprise side, where it is much more fee for service and we are dealing with older gear that clients generally, if you will, sweat the assets longer, it is more of a fee for service, and the margins in there can be more like mid-20s to 30s.
And that is the part of the business that we think inherently will be the larger and longer period of growth in the ALM business for us over time. Not to say that the hyperscale business is not growing rapidly, as you point out. It certainly is and can continue to in light of the TAM growth.
Yes. Okay. Let us talk a little bit about the records management business, both the physical and the digital solutions part. So the physical part has seen volumes be relatively stable for a long time now, but flat to up. And you are seeing customer consolidation, you are seeing outsourcing opportunities. How durable would you say that volume growth profile is in that legacy physical records management business, and what could cause the growth to move above or below the current range, which is somewhere between zero to 50 basis points of growth?
Yeah. So for people who have heard me over the years, I have been with the company now almost seven years, I kind of keep saying the same thing about that part of the business, George. We love that business because it is very cash generative and it is super durable. We have been growing the volume, the physical volume in that business on an organic basis for many, many quarters now in a row, and really it's years at this point.
Yeah.
Together with strong pricing. This year, last year, we were actualizing around 6% pricing benefit. The volume is growing, albeit very slowly. Like we've always said, we expect it to be slightly up, think like 20, 30, 40 basis points a year. With that model, that business generates a tremendous amount of cash.
It speaks to the fact that how we doing all that, how we doing the pricing, the revenue management actions together with incremental volumes, because we're giving our clients a really good value. We continue to bring out services that they can't get from any other vendor in this place. There's plenty of competition.
We offer a worldwide solution for large corporates, and we've got a very consistent track record of doing this in a highly secure and effective way. We make it very easy for them to do business with us. We expect that business to continue to be able to slightly up on a volume basis for the foreseeable future, together with strong pricing. As you note, there are some markets, India is the one I talk about the most frequently because it is the largest opportunity for us, where outsourcing is just really starting as it relates to records management. There's a huge opportunity for additional growth there, and the margins in India are very good as they are around our records business across the globe.
We think that can continue to grow at a mid-single digit plus clip rate for a long period of time. Then you started to ask a little bit about the digital solutions business.
That's a business that is actualizing on an annual basis today at north of $600 million of revenue. You don't have to go back many years where it was sub-$200 million. Back five or six years ago, it was principally a scanning business and very project-oriented, meaning we had to win business each year to keep going. But now the team has successfully been winning more and more business that's recurring and that is multi-year in nature. So we start each year with a larger base of revenue generation, and we're getting into many more software-as-a-service deployments utilizing our DXP technology, which enables us to help clients with content that they have historically not been able to use, essentially dark data. With various new AI tools, together with our DXP platform, we think our digital business has got a lot of opportunity for additional growth, George.
Mm-hmm. Sticking with digital, how much opportunity do you think the new contract with the Treasury Department will contribute to revenues this year and next year, and how much of the growth will be driven by that versus traditional DXP deployments?
Yeah. That's a good size contract, and as many people have probably heard me talk about in the past, it was a complicated procurement situation over the prior year. This year, now that we're kind of doing the work for the government, we expected to do at the beginning of the year, I said we'd probably do about $45 million of revenue. We were a touch ahead of our expectations in the first quarter on revenue, and we were a couple of 2, 3 million ahead in the second quarter. We will probably exceed that level that was initially expecting. We have noted multiple times that next year and for the years beyond, we expect to do at least $100 million of revenue on that business. The government did award the contract to four potential partners for that.
We were the largest player, and I think we're doing much more business than any of the other players. As our service level agreements that we have with the government, we are doing in a very diligent way and meeting all of the requirements and then some that are under that contract, because as you would imagine, the government is very sensitive to the confidentiality of that sort of work and the chain of custody, and frankly, speed and quality and efficacy of the work.
Our teams are doing a great job there. If we got 100% of the theoretical volume that the government put out to bid, we would probably generate about $150 million of revenue in any given year, George. Our expectation was in light of just having multiple vendors, et cetera, we think we'll get the majority of the market share. One of the things that is a factor is, of course, both how many returns are coming in, how much correspondence there are with taxpayers, but of course also how quickly the government outsources the business.
That is one of the reasons why this year we anticipated $45 million. As I mentioned, we are running a little ahead of that. Then ramping over the next year plus, as the government continues to essentially learn how to outsource what is something that they have always done internally.
Yep. Makes sense. I have got a couple more questions, but I am going to pause here to see if there are any questions from the group here. No? Okay. Let us move on to margins and cash flows. The ALM business is a bit margin dilutive to overall company EBITDA margins. Can you talk a little bit about what your outlook is for consolidated margins over the next couple of years, and what factors could drive reported margins to expand?
Yeah. The way I usually explain this is a couple of things. If you look at our business, we have five or six distinct businesses that make up Iron Mountain. So there is the records management box business, which is our highest margin business, and it is doing very well. Then you have got box services. Those are professional services that go along with the box. You have got our digital business we were just talking about.
Our data center business, and the two different ALM businesses, the hyperscale and the enterprise. The enterprise margin, therefore, is the amalgamation of all of those together. As you pointed out earlier, the ALM business is lower margin. The enterprise business, as I mentioned earlier, is more like 20s to 30s, whereas the hyperscale is lower than that. If those businesses are growing faster than the rest, then naturally that is just an obvious headwind to the actual enterprise margin. The thing that is really important to note is, in each one of those businesses I just highlighted, they are all increasing their margin within their businesses. They have the opportunity, and we see the ability over the next few years to continue to increase the margin across each one of those businesses. It just is a relative growth rate.
When I think about our two biggest growth businesses in the growth portfolio you were mentioning earlier, in one case you have data center and you have the other, which is ALM, and they are both this year going to be a little right around $1 billion of revenue. They have both got huge opportunities for growth.
Data center is an accretive margin for us. We are doing like low 50s EBITDA margins in data center today. As that continues to grow and lift its margin, and as ALM grows and we lift the margin within ALM, there is a lot of opportunity here, George.
Right.
The other thing is, while you point out ALM is kind of a dilutive margin to the total, I will just say it is all incremental to the company. It is not like it is taking from any other part of the business. That is a truly incremental business for us, and it is a franchise which we think can be measured, as I said earlier, in billions with an increasing margin. We are feeling really good. Furthermore, the ROIC on that business is actually quite high relative because there is just not a lot of capital required to grow it.
Mm-hmm. On the topic of capital requirements, a lot of the CapEx spend today is being deployed to data centers. As you think about future growth investments, including data centers and other areas of the business, what are your expectations for cash flow generation, cash flow conversion? How should that evolve from where it is today?
Yeah. We're kind of an operating company inside of a REIT structure, as you know. Unlike a lot of REITs, we generate a tremendous amount of cash off of our core businesses, which we then can use to fund portions of our growth. If you look at it, George Tong, I would say our cash available for discretionary items is growing appreciably, think like hundreds of millions each year over the next few. That means that we can continue to build out our data center portfolio with relatively less debt each year.
We have a dividend, and our framework for our dividend is that we're going to pay out at a low 60% of AFFO. If you look at that as a result of how fast we've been growing AFFO over the last few years, we've grown the dividend, call it 10%, each of the last few years. With continued very positive outlooks for AFFO, you should anticipate the dividend will continue to rise. But that works very well with the growth of the free cash flow off the core, together with the fact that while we've got a lot of megawatts to build out, as I said, we're not a speculative builder, so we're going to build those as they energize, hopefully with pre-lease contracts.
As a result, the incremental call on the capital is pretty consistent in here as it relates to how much we are spending on data centers over the next couple of years, subject to additional land purchases. We think we can continue to run the business at leverage levels right in here of, say, we are a little below five times which is our target, midpoint of our range. Generate increasing levels of free cash flow to invest in the growth.
Wonderful. We are just about at time, Barry. Thank you for the great discussion.
Thanks, George.