Data Centre Titans: Building AI Infrastructure Needs a New Delivery Model – Tom Babbington | Adaptive MDC

AI infrastructure demand is placing new pressure on data centre delivery models.

In this Data Centre Titans interview, Tom Babbingtton, Founder and CEO of Adaptive MDC, discusses how integrated delivery, modular construction and clearer accountability can help operators bring AI-ready capacity online more effectively.

– – – – – –

What is your role today, and how did you arrive at this point in your career?

I’m Founder and CEO of Adaptive MDC, a modular data centre delivery company we officially launched at Cannes in June 2026. The route here wasn’t direct. I’ve spent 25 years in construction, the last 15 of them dedicated to data centres, across three major roles with general contractors delivering DC projects in Europe and the US.

Founding Adaptive was the next logical step after two and a half decades of watching the same structural problem repeat itself, project after project: fragmented delivery chains, misaligned incentives between designer, procurer and builder, and no single party accountable for the whole outcome. I wanted to build a company that removed that problem structurally, rather than one that just worked around it a little better than the last contractor.

It wasn’t one moment. It was the accumulation of watching the same failure mode repeat itself, project after project, regardless of the client, the contractor or the geography: fragmented accountability, sequential processes, too many parties with partial responsibility, and no single entity that owned the outcome.

Over a long enough career, you see enough programmes overrun, enough clients absorb costs that could have been avoided, enough situations where everyone had a defensible position, and nobody actually owned it. At some point, you stop accepting that as inevitable and start asking what a different structure would look like.

What I kept coming back to was the parallel manufacturing principle: there is no technical reason design, procurement, and site work can’t run concurrently. The reason they don’t is the delivery model, not the technology. Adaptive was founded to build the model that changes that.

Your career has taken you from international development projects to hyperscale data centres. What experiences have most shaped your approach to delivering critical infrastructure?

After university, I ended up in London, restless, before a role with the UNDP in Afghanistan set the course for everything that followed. From there it was Foreign & Commonwealth Office embassy programmes across Kathmandu, Addis Ababa, Tel Aviv, and Havana, then Wembley regeneration and hotel construction across Europe and the Americas, before I moved into hyperscale: Google facilities with ISG in Northern Europe, then seven years with DPR in Amsterdam.

I grew up in a military family, moving constantly, so operating in unfamiliar environments was never something I had to get used to. What that path taught me, in embassies and in hyperscale halls alike, is that infrastructure delivery breaks down in the same place regardless of sector: at the handover points between design, procurement and construction, where accountability gets lost between parties.

What’s shaped my approach most is discipline: being able to coordinate the right mindsets and collaboration across a team when there are that many moving parts to manage. A lot of that comes from the travel and the life, working overseas, operating in a random location and just getting on with the job.

I moved into data centres almost by accident. I didn’t really know what one was until I got the job on a big hyperscale account, but when you look back, you realise it suits you: you already understand how the moving parts fit together and how to stay disciplined somewhere unfamiliar. It’s honestly also one of the reasons the industry struggles to find the right people. It’s a tough lifestyle, away from family, with high expectations, and not everyone is built for it.

The other thing that’s shaped my approach is a refusal to conform just because something has always been done a certain way. A fast-evolving industry means you can’t keep delivering it the way you did last time. Your history helps you make better-informed decisions, but you still have to keep challenging the norm and looking for a way to add more value. An example of that was when I was working on a hotel construction project in Tunisia.

The workforce wasn’t turning up reliably, and paying them in advance just meant they’d take the money and disappear for a week. So, we stopped paying in advance and started cooking them dinner instead, every night, for the whole team. Their families ended up joining too, but the crews stayed until the day’s work was actually done, and productivity improved. It’s a small thing, but it’s stuck with me: you get more from people by understanding what they’re actually missing than by tightening the terms of the contract.

Having worked on major infrastructure projects across more than 40 countries, what lessons have you learned about what makes large-scale projects succeed – or fail?

The lesson that has repeated most consistently is that projects fail less often because of technical problems than because of structural ones: fragmented delivery chains, where design, procurement, manufacture and installation are handled by different parties with misaligned incentives and no single point of accountability. You see enough programmes overrun, enough clients absorb costs that could have been avoided, enough situations where everyone had a defensible position, and eventually you stop treating that as inevitable.

The projects that succeed are the ones with one accountable partner across the full programme, design through commissioning, so there’s no one to point at someone else when something slips. My own values on this were shaped as much by the projects that went wrong as the ones that went right.

How should the industry’s approach to data centre delivery evolve to meet growing AI demand?

The data centre industry knows how to talk about innovation: AI, high-density compute, liquid cooling, faster deployment. The problem has never really been innovation. It’s execution. Multi-year timelines, disconnected contractors, and layered supplier structures have become accepted as normal, even as operators are under mounting pressure to deploy faster than ever.

The industry needs to move from sequential, fragmented delivery to an integrated model where design, procurement, manufacture, installation, and commissioning run under one accountable structure, with manufacturing and site preparation running in parallel rather than one waiting on the other. That’s the only way to close the gap between how fast AI infrastructure demand is arriving and how slowly traditional construction can respond.

What’s driving the growing interest in modular construction and off-site manufacturing, and what opportunities does it create for the industry?

Honestly, I don’t think anything’s fundamentally changed about the industry. The demand to deliver as quickly as possible has always been there. What’s changed is the number of customers asking for that unachievable timeline. Customers have always said they need something online in 12 months, but now it’s not one customer saying that; it’s an exponential number of them, so we have to find alternatives.

We’re an industry full of great engineers coming up with solutions. The harder part has always been putting those ideas into practice; that’s the execution problem. Modular lends itself to that because it lets you standardise a design enough to actually deliver it at speed. But it’s not standard for standard’s sake: you still have to design a solution to meet the customer’s requirements, because nobody buys a data centre off the shelf. The nuance of what goes in it and how it operates always changes, so our approach is to keep the design scope-specific while optimising what’s already available.

Location still drives a lot of it. Some of the biggest modular jobs right now are in the Middle East, because you can control so many more elements of the build there. In smaller markets where concrete is king and the workforce is already in place, traditional methods still make sense. Modular doesn’t solve every problem; however, it’s an alternative that lets you use available resources more efficiently, and there’s plenty of manufacturing capability across the world that can support it.

As demand for data centre capacity grows, where do you see the biggest constraints emerging?

For me, the biggest constraint isn’t really technical, but it’s policy and public perception. Yes, there are real resourcing and capacity issues within the sector. But externally there’s a set of assumptions about what the problem with data centres actually is: that they don’t bring jobs, that they just use power and water. The reality is closer to the opposite. Data centres are becoming an essential part of a productive society, giving people access to the AI tools that let them be more productive and help economies grow. They’re not a commodity driven by a big tech company sitting off to one side. They’re infrastructure responding to consumer demand, and an add-on to growth rather than a drain on it.

The energy sector has had two hundred years to build the policy framework that guarantees everyone a connection to the grid. Data centres haven’t had that runway, and the public understands what power does in a way it doesn’t yet understand what a data centre does. Energy providers, data centre operators and policymakers are still working in silos when what’s needed is to move in tandem. Closing that gap in public and policy understanding is, to me, the biggest constraint on how fast this industry can responsibly grow.

What do you think are the key ingredients for building trust on large-scale infrastructure projects?

Trust in infrastructure isn’t a technical problem; it’s a human one. Reliability, quality, and doing exactly what you said you’d do. That’s it. We work in a service industry, which is often forgotten in a sector so obsessed with engineering. But the moment you start treating a project as a collective goal, rather than a contractual boundary to defend, everything changes.

The businesses that struggle are the ones focused entirely on margin and swim lanes: heads down, contract in hand, asking only “what am I required to deliver?” That mindset kills projects. Nothing in large-scale infrastructure can be done in a silo. Too many moving parts, too many interdependencies. Do those things right – keep the relationship good, pay people on time, get paid on time – and you’re guaranteed to build trust.

That’s what we set Adaptive up to be: a genuine service proposition. Yes, we have design concepts and technology behind us, but the whole point is to give customers the full-service lifecycle: design, delivery, commissioning. Not just a product you buy and hand off to someone else to install. When you own the whole outcome, trust isn’t something you have to build. It’s just what happens.

If you were advising someone entering the data centre industry today, what skills or mindset do you think will matter most over the next decade?

The people who will thrive are the ones who refuse to be defined by a job title. Don’t silo yourself. The moment you think “I’m just here to do this one thing,” you’re already behind.

More than any technical skill, I’d look for someone who feels genuinely rewarded by the work, and I mean that both ways. You have to be giving your best, and you have to be getting something meaningful back. If you feel taken advantage of, you won’t last, and you won’t perform. The best people I’ve worked with are engaged, curious, and willing to take the game to the opposition rather than sit back and react.

The Adaptive journey has shown me that firsthand. The data centre work isn’t new to us.  What’s genuinely new is building a company from scratch, with all the complexity that brings. The teams who will define the next decade of this industry are the ones who lean into that complexity rather than hiding from it.

Looking ahead, what’s one bold prediction you would make about how data centres will be designed, built, or delivered over the next five years?

They’ll be 3D printed. I’ve joked about this for as long as I’ve been in the industry: why don’t we just 3D print the whole lot? Think Rocky in Project Hail Mary, conjuring structures out of exotic materials on demand. We’re miles from that, but I’d love to see it.

More seriously: What I genuinely believe is that data centres will become a call-off commodity. Standardised, factory-manufactured, deployable on demand. The conversation will shift from “can we build this?” to “where do I want it and when?” You’ll specify your capacity requirement, confirm your powered land, and call off ten megawatts of IT infrastructure – fully integrated, fully tested – arriving on site within three months and live within six.  That’s where the industry is going. And that’s the programmatic, at-scale, modular delivery model we’ve built Adaptive to deliver.