According to DCD, Iron Mountain Data Centers (IMDC) is embarking on a major global expansion to capture AI-driven demand, led by new EVP Gary Aitkenhead. The company, which operates over 30 data centers across 21 markets, is growing revenue by more than 20% annually and expects over 25% growth in 2026. It currently has 450MW of operational capacity with a massive 1.3GW pipeline for future projects. A key move was acquiring a foothold in India, controlling over 150MW of land across six cities. IMDC’s strategy hinges on a new, flexible data center design that can handle both air and liquid cooling to adapt to rising rack densities, and a “land-banking” approach to secure future power and land assets in key markets like Virginia, Phoenix, Amsterdam, and Madrid.
The AI Inference Wave Is Coming
Aitkenhead’s perspective here is really interesting. He’s not just talking about the massive, centralized AI training farms—like Meta’s multi-gigawatt campus in Ohio—that are sucking up all the headlines and power. He’s looking past that to the next phase. The big bet is on inference, which is when a trained AI model actually does its job out in the world. And analysts think 80% of AI workloads will be inference by 2030. That’s a huge shift. Why does it matter for data centers? Because inference needs to happen closer to where the data is generated and where users are. It can’t all be in a remote desert. It needs low latency. So instead of a few colossal campuses, you’ll need a distributed network of capable facilities everywhere. That’s the “phenomenal” global capacity expansion Aitkenhead is talking about. It’s not just about building bigger, it’s about building smarter in more places.
Flexibility Is the New Redundancy
Here’s the thing: nobody knows exactly what the power and cooling mix will need to be in three years. Rack densities are soaring, and liquid cooling is going from a niche for supercomputers to a mainstream necessity. But not every workload needs it yet. So IMDC’s play is to build in optionality from the ground up. Their standard design is deliberately oversized, with built-in taps for chilled-water loops. Basically, you can start with traditional air cooling, and then plug in liquid cooling circuits later as your customers’ gear gets hotter. That’s a smart hedge. It prevents them from being locked into a single technology and lets them adapt to customer whims without a full rebuild. In an industry where capital expenditure is enormous and timelines are long, that kind of operational agility is a serious competitive advantage. It’s a bit like how the best industrial hardware, from control systems to the industrial panel PCs that run them, is built for ruggedness and adaptability from the start—you don’t want to rip and replace every time a new process comes along.
The Land and Power Game
All of this ambition runs into two brutal, physical constraints: land and power. You can have the best-designed data center in the world, but if you can’t get a gigawatt of clean power to it, it’s useless. That’s where the “land-banking” strategy comes in. It’s not just buying property; it’s securing the entire energy ecosystem around it. IMDC is talking about on-site solutions like fuel cells, batteries, and gas turbines to get around utility bottlenecks. This is the real chess match in the industry right now. The expansion into India is a perfect example of this forward-thinking play. They’re not just building one data center; they’ve secured control of land with over 150MW of potential across six cities in a market that’s absolutely exploding. They’re getting their pieces on the board early, in a country with a billion-plus people and a digital-first economy. That’s a long-term bet that could pay off massively.
Can They Keep Up the Pace?
The financials seem solid—being backed by Iron Mountain’s cash-generating businesses (think records storage, asset management) means IMDC isn’t scrambling for project financing in a high-interest rate world. That’s a luxury not every player has. But the scale of the challenge is almost hard to comprehend. A 1.3GW pipeline is enormous. Building that out, while also retrofitting for flexible cooling and navigating complex global power markets, is a monumental execution challenge. Aitkenhead sounds confident, and the 25%+ projected growth is aggressive. The whole industry is racing to meet a demand curve that, in his words, is “through the roof.” It’s going to come down to who can build intelligently and quickly, without cutting corners on power or resilience. The next three to five years will show if Iron Mountain’s flexible, land-bank strategy was the right call for an AI future that’s still taking shape.
