Third Bridge专家访谈称Iron Mountain数据中心需求仍强但缺少新增可交易催化

[[00:02:01]]
Iron Mountain reported that Q1 2026 revenues grew 44% YoY in the global data centre business with a slight contraction of the adjusted EBITDA margin by 30bps to 52.1%, although this would have been a 120bps expansion if you exclude power pass-through costs. What are your impressions of the rapidly growing data centre business at the company? How effective do you think it's been at scaling this data centre business?

[[00:02:43]]
I think anyone that's been in the data centre business, let's say, prior to 2017 should be in a great position for success. Obviously, a lot of the success is driven by the demands of the new AI workloads, the learning models that are needed for high-density cooling but also the sites that are up and running. I think Iron Mountain has been successful with expansion and upgrades needed for some of the campuses where they've always had a strong presence. Specific would be Phoenix, Arizona and New Jersey come to mind, but more recently, a big part of that success, I would have to guess, is attributed to the expansions and the new client acquisitions that they've had in Manassas, Virginia. That's, let's say, south of Ashburn, where a lot of people with data centres are familiar with. Iron Mountain did a good job with that campus in Manassas.

[[00:03:58]]
Iron Mountain's development strategy seems to rely heavily on large-scale campuses, such as Richmond, which has 200MW-plus in capacity and is expected to energise by 2027. Other campuses include London and Madrid. What was the core developmental strategy for the data centre business during your time at the company?

[[00:04:32]]
During my time with the organisation, part of my job was new client acquisition but also looking at, let's call it, Iron Mountain's traditional solutions and customer base, which was predominantly enterprise. I would define enterprise as the highly regulated companies within the banking, healthcare industries. We did have a team focused on the federal government here in the US. They also had a significant amount of success in regions as well. I had peers on the data centre side that focused on that. What I thought was interesting in terms of where Iron Mountain was positioned from the time I was there was that everyone spoke about the transition away from paper or traditional media. Everyone thinks of Iron Mountain as the paper shredding company or the tape backups or the tape destruction or the hardware destruction company. Even during my time and perhaps arguably now, no one realises how mature and sophisticated their data centre solutions were.

We were a household name within, let's say, disaster recovery, business continuity, chain of custody but not so much of data centres. I think since then, they've been able to capitalise on success with making the right investments in the markets that you described, Manassas, Virginia, Richmond coming online in the coming years, Madrid. India was a big focus while I was there as well. I think they're finally now the sole owner of that acquisition from Web Werks, so we'll see. I would expect success in those markets for that. In terms of what brings the success, it goes along with the Iron Mountain name. Again, there's not too many people that have been doing business for as long as they've been doing it. I also think just the executive team has done a great job hiring the right folks and working with the right individuals to capitalise on the right assets, so finding a market like Manassas years ago before it was saturated and being able to have plans to develop in the right moment with the right customer base.

[[00:06:55]]
How much of Iron Mountain's global data centre business is comprised of enterprise customers vs hyperscalers, if we look at the total megawatt capacity across these facilities? What does the profile mix look like between those two customer cohorts?

[[00:07:32]]
I'm hesitant to say that, hey, listen, this is what it is for sure. Generally speaking, I would say that newer builds, without a doubt, are going to have the majority of capacity, let's say 70-80% of that capacity and that's in terms of power that's being delivered to, let's call it, the hyperscale neocloud environments. While some of their legacy sites, let's say, for example, AZP-1 and 2 in Phoenix, their Scottsdale site, let's say, a big portion of their New Jersey site, Denver, those remain colocation sites where you'll have dozens if not hundreds of customers within there that they'll continue to manage and be successful with. Again, generally speaking, I think the strategy for colocation providers, not just Iron Mountain but anything new that's being built, I think that the ideal target these days are going to be a large customer to assume the majority of that capacity that's coming online.

[[00:08:46]]
Can you compare and contrast the overlaps and differences between enterprise customers and hyperscalers in the sales process or the customer profile in general? I tend to think a hyperscaler tenant is going to be able to command more favourable pricing, so margins might be lower with that tenant, but you probably have more of a guarantee of it sticking around. Conversely, margins might be higher with enterprises, and you're probably looking at 100-200kW to around 5MW at the most in terms of overall workload deployment or footprint. That footprint or deployment size is much larger with hyperscalers. Could you articulate some of the differences and overlaps between these two customer cohorts?

[[00:09:43]]
Is there a time period in which you want me to focus or comment on? Are you referring to more recently in the last two or three years or even in general?

[[00:09:53]]
Perhaps over the last three years?

[[00:10:05]]
If you ask anyone in the industry, let's go back to 2019 because I do think the 2020 pandemic is when things accelerated at a rapid rate. 2019, you had your hyperscale data centres. It's safe to say that anything above 2MW was considered a large-sized deal, more than likely going to a technology company. Iron Mountain's sales force and our competitors out there, we were focused on the enterprise opportunities. Again, traditionally, one project, 2MW or less, but you would always expect to be successful. You'd find an opportunity in the 500kW-1.5MW or, let's say, 1-2MW range and everyone would be thrilled with that. What we started to see after 2020 was the need for data centres to accelerate rapidly. We started to get a sniff of workloads being more high dense. Some of the sites were built with an 8-12kW per rack design, but now, companies started asking for, let's say, 20kW to as much as 30kW per rack. That's a time period of, let's say, 2019-20, where you started to see those higher dense workloads supporting AI models.

Nothing crazy yet, but I think it was around the time Russia invaded Ukraine when energy prices spiked and the requirements started becoming not crazy, but everyone started trying to find land and providers where they could secure significant amounts of power. I think traditionally, before that, hyperscalers were ones that were driving price down. They would enter a relatively new market or a tier 2 city and they were asking for pricing of, let's say, USD 100 per kilowatt or USD 120 per kilowatt. Around that time, something changed where the developers and providers started to see that there was far more demand than what we could finally keep up with. We had to adjust our pricing because of the increase in energy costs and because of inflation, because of how much it was costing to build the data centres. I remember there was a deal, not just that I worked on but others in the industry, where we started to see that number go from, let's say, USD 100 per kilowatt all the way up to USD 150-160 per kilowatt and the hyperscalers were agreeing to this, especially if it was significant capacity down the line.

I think that changed where the industry was going because hyperscale was doing a job of driving the cost down. In this case, it was going up. I thought during that time, that's where I saw a challenge with balancing who the profile customers for, let's say, colo companies are. Hyperscales were ready to secure space and power at those rates, but enterprise certainly wasn't. There were a number of deals that I was familiar with where negotiations were going on for 6-10 months or 18 months. Customers traditionally expected those numbers to be secure, but because of how the industry changed, because of inflation, we had to take back some of the pricing shared and then readjust to those figures that were, let's say, anywhere from 20% to 40% more than what was originally negotiated. Some enterprise were okay with that. Some decided to put holds on those projects. I thought that was interesting. In terms of architecture, service delivery, hyperscalers traditionally, they're looking at environments at a minimum 20MW with the new AI workloads. If you were talking about liquid cooling, that's always a plus or availability of liquid to the racks is always a plus to support those next-generation workloads. Then those are going to be in the range of, I would say, at a minimum 20kW per cabinet but up to 60kW per cabinet that should be enabled for enterprise.

Now, they're dabbling in that, especially the larger companies but still not growing at the pace that hyperscalers and let's call them the neoclouds are growing at. One aspect of the business that I'll comment on is, again, speaking in general terms, what a data centre looked like in, let's say, 2019-23. That was a race to turn up capacity at medium densities. A lot of these sites were built on slab. A lot of the sites used air fans and didn't use water to the rack because they want to get out to the market quicker. After '22 or '23, it's funny how things change, everyone needed water to the rack and then raised floor was beneficial because if you were retrofitting a site to have water to the rack, you could keep the water underneath the raised floor, run piping underneath for delivery to those environments. It was an interesting time because not just Iron Mountain but in general, everyone had this capacity turning up, but there are limitations on how it could support the AI workloads that were coming through. Now, what we're seeing is a change in that architecture, considerations being made of air-cooled for time to market, but the ability to deliver water in the event that water to the rack cooling is needed.

I went in a lot of different directions there, but hopefully, I was able to touch on some of the aspects of the hyperscale and neoclouds on the contrary than the enterprise environments.

[[00:18:08]]
There's been a lot of back and forth over data centres and the general AI build-out over the course of 2026. It was happening earlier, but really accelerated in 2026. We talked about EBITDA margins in Iron Mountain's data centre business – 150bps improvement in Q1 2026, if you exclude power pass-through costs that made it a 30bps contraction. One way to read that, which came through in management's comments, is that the headwind arose from stepped-up power consumption from tenants and customers for new cloud and AI deployments that have been ramping up faster than usual. That amplified data centre power consumption really speaks to the strong underlying demand, both for power and data centre capacity, so Iron Mountain's customers, but we have the contrasting viewpoint as well. What is your perspective on the underlying demand picture? Are you seeing strong demand for power and data centre capacity, or has that been peeling back?

[[00:19:06]]
Man, the amount of growth we've seen in the last two years in terms of colo providers acquiring land, hiring to work with developers, construction companies and meet the demands of what they're seeing has been unprecedented. In terms of where we are today, are we slowing down? There might be a little bit of a slowdown, but relative to where we were, let's say, two years ago, I wouldn't call this a slowdown period. I think the industry is going to continue to have demands. I think the only risk that I would be aware of, and this isn't specific to Iron Mountain by any means, this is just the industry in general, is the politics that are going on. You mentioned 2026. I think those of us in the industry could agree, this is the first year that data centres truly is mainstream. When you go on community message boards like Nextdoor, Facebook, etc, almost every single community has a topic of data centres that seems to have a large thread. I would describe some of it as panic hysteria, but that could be my biased point of view. I think that being said, combine that with the politics and what you're seeing in my own state, New Jersey, and where I grew up in New York, I think that would be the biggest risk to the industry.

I also want to believe that the technology community, the hyperscalers, the industry professionals, the developers, the construction companies, are innovating and coming together to address the concerns and calm the tension with communities, so that people understand those benefits and the growth will continue. I think the industry will continue to go. I think people's appetite for data consumption will continue to grow. As a result, data centres are in a good place in terms of investment. Whether or not let's say that that piece of the pie has to be given back to communities or the cost of building will be more expensive because of raised concerns and standards to the environment that people are expecting, I think that could certainly put a dent in it, but I don't see things going away. I think if they do, in my personal opinion, that would be a major concern because AI is a race that other countries are fighting for. The US without a doubt today is ahead of that race. It wouldn't be in our best interest to respond to some of the panic and hysteria that we're in the industry that we're seeing.

[[00:22:20]]
We're seeing strong underlying demand. Earlier, you talked about increasing rack density requirements, which we've seen growing over the last few years. Iron Mountain's full-year 2026 guidance is calling for USD 2.05bn of growth CAPEX and USD 150m of maintenance CAPEX, slightly down from 2025 levels on a total CAPEX basis. Given these increasing rack density requirements, how much of that CAPEX will have to be allocated to retrofitting existing facilities to accommodate higher density cloud and AI deployments that we're seeing across hyperscalers and even in enterprise tenants?

[[00:23:23]]
It would be tough to give you a percentage. If you look at their portfolio in terms of where they have sites today and then you take a look at what's expected to be newly built, I think a significant amount of their growth and if we call growth, a percentage of power that they're making available will be these newer architectures and environments to attract hyperscale customers and neocloud customers. In terms of retrofitting, I don't think it's going to be a lot. New Jersey is a site that comes to mind, but even then, whatever New Jersey has, let's say, capacity-wise, if they were able to grow, let's say, 20MW within that market and let's call that 20MW of retrofit, that's still a fraction of power of what they're anticipating in terms of growth in Richmond, in Manassas, in Chicago, in Spain for that matter in terms of new capacity. I don't see a significant amount in terms of retrofitting for now just because those sites, the ability to do so is still a fraction of power of what they're seeing on the growth side.

[[00:24:46]]
Iron Mountain management has guided for over 100MW of new data centre leasing in full-year 2026. Thus far, it's been slower than expected, with a reported 22MW signed in Q1 and 10MW in April. We saw a similar set-up in 2025 where initial guidance called for greater than 125MW, which was eventually cut to 30-80MW and closed ultimately at 60MW. How realistic do you consider the full-year target of over 100MW?

[[00:25:57]]
I don't know specifics of how successful they may be or how likely they are to hit their targets that were presented over the last few months, but to what you asked in terms of my experience of the company and what I saw, I thought Iron Mountain did a great job of hitting those targets and being successful. I joke when I started at the company, I think their stock price was at USD 30 per share. Then by the time I left, it was over USD 120 per share. I think they've done a good job of managing expectations and being successful there. If that's what they said they would do, again, based on my experience, I think they do a good job of setting the right expectations and being successful with those targets.

[[00:26:51]]
You mentioned some other colocation operators. Iron Mountain is not as well-known of a name in this area, but it is with existing customers, particularly enterprise, RIM [records and information management]-type account customers. It presents this massive potential cross-sell base for the company to tap into for colocation services, whether that be financial or healthcare verticals. How effectively did you find the sales organisation in terms of targeting and tapping into that existing enterprise RIM base?

[[00:27:46]]
When you say targeting those enterprise new logos, are you saying a traditional customer and then getting them to take colocation data centre space with us?

[[00:27:55]]
Yes, exactly. That might have not ever been the priority in the first place, of course.

[[00:28:05]]
When I was hired at the organisation and even at the time I left, there was always, let's call it, a KPI for salespeople to work with our peers on what was said to be the other side of the mountain. We always want to take customers that were spending millions with us on the traditional solutions and qualify interest in the data centre solutions because, quite frankly, there's better margins on that side of the business. I would say the teams were successful on, I'd say, a good amount of deals, but by no means was any of the success on the data centres dependent on the incumbent traditional customers that we had. I think that's worth distinguishing and calling out. I'll also say that part of the challenge, though, especially in the, let's say, 2023, part of '22, and again, not just with Iron Mount but in general, was the competition to secure capacity. As I mentioned earlier, hyperscalers were willing to pay premiums to secure a good chunk of new builds that were taking place. Enterprise wasn't. You couldn't count on a hospital or you couldn't count on a large bank to say, "Hey, we might need this in two years, so we will sign an agreement with you at a premium price." There were deals that were put on indefinite hold.

Again, with the buying habits of the customer profile changing, especially in 2022-23, enterprise projects where the decision-making could take 9-18 months, that was the norm, at least in the years I've been doing the business. Those habits changed. Hyperscalers forced the hand where if you had capacity, they were willing to take it. That would essentially take away from what was originally, I don't want to say earmarked for the enterprise but expected to sell to the enterprise. That's where a lot of activity was different. I joke around in my career, we're always competing for business on the enterprise side. You're lowering price, you're being flexible with terms, but something happened after the pandemic where instead of a buyer's market, it certainly became a seller's market.

[[00:31:15]]
We talked about the huge investment Iron Mountain has made into Richmond, which has 200MW expected to be energised by 2027. I don't think we've seen any pre-leasing commitments as of yet, but the idea is to get in large hyperscale anchor tenants. How do you think that capacity needs to be made more attractive to hyperscale customers over the coming months, or is that not an issue? Is it more that both parties are trying to hammer out design specs and be diligent around power concerns and contract terms? I imagine the company is active in discussions with all of these parties right now. Can anything be done to make the Richmond space more attractive or is it already attractive enough?

[[00:32:23]]
I'm not aware of any issues or that being an asset to be worried about. I don't know specifics in terms of who they are talking with or pre-selling or who they have commitments with. I can tell you that anyone with that much land secure with the permits in place that give a roadmap of energisation and turning on capacity, I think you're in a great position to attract a handful of customers to assume that capacity. I think they're in great shape for that. Again, I'm not aware of specifics, but I would have to imagine that it's only a matter of time before they may announce that most of that space has already been pre-leased or secured.

[[00:33:15]]
I believe Iron Mountain has noted that capacity in Richmond isn't even leasable until H2 2026, so volumes could be more back-half weighted.

[[00:33:32]]
Usually, the way that the conversations would work is we're speaking to the customers or they're speaking to new customers that here, there's expansions going within certain markets. We share what that pipeline looks like. We share what the timelines of that look like. If they're serious, then they start talking about designs. It could be, "Hey, here's what we have in mind. This will be air-cooled but liquid-ready if you need it. Targeted density, targeted PUE is going to be this." Then if those timelines fit into theirs, then they'll start talking about how to build this to their spec or as close to their spec as possible.

[[00:34:21]]
I believe the build-to-suite approach really resonates with hyperscalers. Was there a general PUE [power usage effectiveness] that hyperscale customers were looking for? Would they more often than not gravitate towards water cooling as opposed to traditional air cooling? What are some of the unique characteristics that you found in negotiations with hyperscalers?

[[00:34:54]]
Having done this since 2008, I have an answer that's a little bit different and maybe even out of context or dated. I look at PUE and I would argue is if I were the one purchasing a data centre or if I was the one negotiating a lease with a provider, my focus for PUE is less on the operations side and efficiency but more so in understanding how those numbers bake into revenue generation for the provider. In other words, what I'm saying by that is, yes, PUE is an operational metric, but at the end of the day, it's also a percentage that the companies are making based on how much power the customers are consuming. In other words, a genuine PUE from an operations standpoint could be as little as 1.2-1.3. At the end of the day, the lease agreement with the customer may be as high as 1.4-1.5. Then that's an understanding that the providers are making a margin on consumption that customers have.

At the end of the day, without a doubt, there's more pressure to operate more efficiently. I think when you're signing agreements at the scale of 20MW, 30MW, 50MW, a company that's making 5% on your power bill, that's a significant amount of revenue. I think that's something that for years, the hyperscalers has been aware of. It's a matter of everyone wants to report a low PUE and high efficiency. In terms of the customers, it is a negotiation in terms of what that means and the margins that are associated with that.

[[00:37:05]]
You talked about Iron Mountain being a lesser-known name in the colocation space. How do you see the company being differentiated from pure-play data centre developers, such as Digital Realty and Equinix, or private operators, such as QTS, Vantage or CyrusOne, all of whom are competing fiercely for that tier 1 market capacity? How does Iron Mountain differentiate itself in the data centre business in terms of trying to win either hyperscale or enterprise business?

[[00:37:51]]
I think Iron Mountain is in a very unique position because of their existing enterprise customer profile even outside of the data centre space. I think in terms of risk, I don't see Iron Mountain as being overly leveraged to some other providers out there because again, they have a commitment in enterprise. That's something that they're unapologetic for on the traditional business side. All those things being said, one would say that's why they haven't grown as rapidly or as quick as some of others in the industry. When I joined Iron Mountain in 2020 and even before that, you look at the portfolios of CyrusOne, QTS, CoreSite, I think you would say that Iron Mountain was right on par with them in terms of cities, customer profile, capacity.

A lot has changed since then. You had the private equity go into those other companies. QTS is now probably one of the largest data centre providers, developers in the world right now with the backing that they have. CyrusOne also a lot of build-to-suite sites that they have throughout their portfolio. One would say that Iron Mountain hasn't grown as aggressively as their industry peers. That being said, if you are of the belief that there is risk with these AI companies, if you're of the belief that there's a possibility of a bubble or demand disappearing, then I think Iron Mountain is in a great position, let's say, from a risk point of view because they haven't built beyond the reach that other companies may be at risk for.

[[00:39:56]]
Iron Mountain's projected timeframe for energising 400MW is the next 24 months, but it's mentioned that that's conservative, probably relative to the broader industry. From your perspective, what is the risk that the company sees slippage on that 400MW target, particularly if we consider several of the gating factors that are taking place across the industry? You talked about community pushback concerns, and we're seeing that rise more in 2026 than any other period before. New York has put a moratorium on all new data centre build-outs, and we're seeing power constraints as a constant. It seems Iron Mountain is relatively positive in terms of the conversations it has with utilities and power partners. We're also seeing extended lead times with supply chain partners to get the infrastructure or hardware itself, whether that's an on-site generator, UPS [uninterruptible power supply] system, etc. How realistic is the target of 400MW over the next 24 months?

[[00:41:38]]
I think generally, that's an achievable number for a company like Iron Mountain. I don't think they're overextending themselves with that. I think there's folks there that are taking consideration of supply chain constraints, lead times, long lead items. Again, I think the biggest risk here in terms of what can change overnight is the political environment in terms of what's going on in communities and the pushback for licences and permits. It's a risk that it makes the cost of doing business far over the budget of what's anticipated. Then in effect, is it worth a customer? Are you able to pass those additional costs to the customer? Is the customer still willing to move forward with a project if whatever is budgeted or forecasted was X, but now it's X times 1.3-1.4, whatever it is?

Again, I think numbers are reasonable for Iron Mountain to meet and be successful with the caveat being the political environment with data centres. I think that's the biggest risk of the industry. Again, you're talking about a two-year, three-year forecast. I think long term industry-wide, so not just with Iron Mountain, if the growth in demand continues on this trajectory, what's being talked about a lot is the skilled trades gap between what's out there and how many people it takes to build and deliver these solutions. In the interim in the short term, a couple of years down the line, most companies are in a good position to meet those targets, especially Iron Mountain.

[[00:43:32]]
Iron Mountain cited USD 2.5bn in growth CAPEX and USD 150m in maintenance CAPEX for 2026. Do you think the company's development CAPEX and general capital allocation strategy are tilting less aggressive? Do you think Iron Mountain was conservative in the first place or was there ever a concern that it might be leveraging itself or getting out in front of its skis?

[[00:44:09]]
I think they're in pretty good shape there. Let's see here some baseline numbers. Then you said the forecast was 200MW or so.

[[00:44:22]]
Yes, I believe 400MW over the next 24 months.

[[00:44:24]]
The numbers seem to be in line because you're looking at anywhere from… One of the baseline figures we use in the industry is anywhere from USD 10m to USD 15m per megawatt. That's right in line with what I would anticipate given those numbers you're sharing.

[[00:44:50]]
I've heard USD 10m-15m per megawatt before. Have you seen that edge higher over the last 12-18 months as a result of increasing costs in construction or more difficulty in securing power?

[[00:45:15]]
Not yet. The number that I've heard, USD 15m, is as high as I've heard in the last six months. I said USD 15m in a meeting recently and someone was like, "All things considered, we're still seeing sites be successful at the USD 10m mark." I think it's safe to say that's still the number that most people would be comfortable writing on a napkin, if you will.

[[00:45:49]]
We were talking about the discrepancies between enterprise retail colocation customers and hyperscalers. Would it be fair to say that there could be a 20-40% price discount on the rental rate you're able to charge per kilowatt hour to a hyperscaler? I believe hyperscalers could command a discounted rate vs enterprise, given how much space they take and the way they can drive the conversation more than an enterprise tenant can.

[[00:46:45]]
Yes, yes. The fact is that hyperscalers are de-risking the development of so many of these sites. For years, I use the analogy of shopping malls that I would assume you and I grew up going to in the '80s and '90s, where you had your large department stores and you had your small retail shops. That's essentially what a colocation environment is. You have your anchor tenants. Instead of Macy's, Sears, Bloomingdale's, you have your hyperscalers. Then within that, you have your retails like the Foot Locker and Gap. Instead of that, you have your enterprise customers like the large banks or healthcare companies. No doubt that they're commanding more aggressive rates. I think what they're also doing is helping de-risk these projects for the developers and for the data centre owners.

[[00:47:41]]
Q1 2026 saw 22MW of new leases and a renewal of 7MW, where renewal pricing spreads were up 12% and 14% on a cash and GAAP [generally accepted accounting principles] basis, respectively, which just goes to show the positive pricing dynamic and power that the Iron Mountain data centre business has right now. Would you agree with that? Do you view the company's pricing power as sustainable going forwards? Is it largely attributable to the lack of supply across the market today?

[[00:48:38]]
I see that being consistent with what's moving forward. I do think it's a response to the current supply vs the demand. We're seeing that across all aspects of the industry right now. The ongoing thing that I've seen even not being on the colo side any more, anyone with 4-7MW of capacity today in North America, you could have about a dozen customers line up to sign those agreements if you have it ready in Q4 or Q1 of '27. I think the demand would persist on what they have. Pricing certainly remains at, let's say, a seller's market right now.

[[00:49:34]]
Iron Mountain's data centre business made up roughly 13% of Q1 2026's consolidated revenue. The company has been funding the business through its traditional global RIM business, reinvesting the traditional storage business into this data centre business and ALM [asset lifecycle management]. How do you expect the data centre business to continue scaling? If it's 13% now, what might it look like at the end of FY27 or FY28?

[[00:50:57]]
Whatever the percentage is now, I think in terms of what success looks like for the company, the data centre business should continue to be the driving force behind its growth and should continue to earn a bigger piece of the pie of all the revenue generated for the company.

[[00:51:18]]
Was it the North Star? What was your impression during your time at Iron Mountain?

[[00:51:23]]
Without a doubt. All of us, we lean into these conversations with Iron Mountain. We are Iron Mountain Data Centers. Our focus is on the data centres that are being built, the multi-tenant owned and operated sites that are being built and we are the fastest-growing business for Iron Mountain. I think almost every person within that organisation, especially the data centre side, would say that unapologetically.

[[00:51:54]]
What are your impressions of Iron Mountain's CEO, William Meany, and the broader management team?

[[00:52:06]]
I think Bill is doing a great job. As far as I remember, Mark Kidd is still leading the charge with the data centre group. He's well-liked throughout the organisation. I know that there's been some changes with respect to, let's say, the construction development aspect of the organisation over the last few years. You're going to see that in any competitive space. We're at a point where everyone is looking to hire anyone. I joke around like you'd be foolish if you didn't take a phone call that someone said, "Hey, we love what you're doing. We're interested in hiring there." I think Bill has been a great leader for the organisation, consistent for the last few years, same as Mark Kidd and the culture that he continues to instil and, let's say, the responsibilities and power he gives to the regions. You see a lot of stability with the leadership teams in North America and Europe. The operations team, I see it today, I think the operations team does a fantastic job managing the sites and providing that service to their customers. I think it's all good from the top down, in my opinion.

[[00:53:27]]
If you had to select an international expansion market that excites you most for Iron Mountain, which would that be across Madrid, London, India, etc?

[[00:53:49]]
It's tough to say right now. I think Southeast Asia is the new frontier. I think geopolitically, it makes sense to enable some of those regions if you get over some of the cultural and risk factors with their infrastructure. In Europe, I think the Nordics will continue to be interesting. I think there's areas in eastern Europe that may be business-friendly. Again, geopolitically, there could be risk there. I couldn't comment with confidence where I would place my USD 1bn bet. Again, I think at the end of the day, if you're able to find that land, find that reliable power, I think anyone with USD 1bn is going to be landing somewhere and marketing themselves as a data centre provider.

[[00:55:04]]
With that, we will end the Interview. Let me close by saying thank you for not only taking the time to speak with us today but for the wonderful input and insights. Thank you clients as always for choosing Third Bridge. We do hope you enjoyed today's content. Lastly, a big thank you to all the team members who have put this discussion together today.

Transcription ends at 00:55:19 of the recorded material