Sam Stacey, director of A7C and acumen7 member, highlights how data centres reveal the need for transformation in the construction industry
The rapid growth of data centres is often framed as a familiar construction challenge: constrained supply chains, long lead times, and pressure on specialist capacity. But that framing misses the bigger picture.
Data centres are not just another asset class. They are the physical foundation of the AI revolution, and in that role, they expose something far more important than a temporary bottleneck. They reveal how the entire construction system behaves under pressure and how it must evolve.
A system of constraints, not a single bottleneck
At the heart of the AI revolution sits a simple reality: it is not limited by one constraint, but by many. The accompanying graphic, which shows how the AI revolution depends on a system of interdependent constraints, illustrates this clearly. AI capability depends on a system of interdependent constraints, including compute hardware, energy generation, data centre infrastructure, cooling, skills, data, supply chains, governance and capital. Each of these sits in a different sector. Each operates on different timescales. Each is controlled by different actors. And critically: A weakness in any one of them limits the performance of the whole system.
Not all constraints are equal
Analysis of these constraints shows an important distinction. Some are structural and slow-moving: semiconductor manufacturing, energy infrastructure and global supply chains. These are extremely difficult to change and operate over decades. Others are more immediate and actionable: data centre delivery, cooling systems and workforce capability. These sit much closer to industry and to construction. This leads to a crucial insight: The hardest constraints in the AI system sit outside construction. But the most solvable ones sit within it.
We haven’t solved the bottleneck; we’ve moved it upstream
The most visible constraint today is in high-voltage electrical equipment: transformers, switchgear and associated infrastructure. In response, hyperscalers such as Amazon Web Services, Microsoft and Google have adopted owner-furnished, contractor-installed (OFCI) models, procuring critical equipment directly from global manufacturers. This has been effective, but it is perhaps misunderstood. The constraint remains. It has just shifted from the construction supply chain into the procurement systems of global technology firms, negotiating for limited manufacturing capacity.
What construction still owns
Even with OFCI in place, the majority of delivery risk remains firmly within the construction sector. Data centres are among the most complex assets ever delivered. The challenge is not simply to procure components, but to integrate them into a functioning system. That includes coordinating complex M&E interfaces, managing constrained logistics and sequencing, integrating multiple suppliers, and commissioning and validating performance. In other words, the problem is no longer just getting the kit; it is making the system work. This is where the construction sector retains both responsibility and opportunity.
The constraint that hasn’t moved: Electricity
There is, however, one constraint that has not shifted at all: electricity supply. Across the UK, grid connections can take 5–10 years, local capacity is constrained, and infrastructure upgrades are slow. At the same time, the same supply chains required for data centres are needed to upgrade the grid itself. This creates direct competition between data centres, utilities and the wider energy transition. Access to power, not construction capacity, is increasingly the gating factor. And yet electricity is still treated as external to construction, rather than integrated into it.
A system under strain
What the graphic and analysis together show is that the AI system is not failing due to a single difficult constraint. It is struggling because of fragmentation. Each constraint is owned by different actors, managed in isolation and optimised locally. But performance depends on how they work together. This is the core issue: The failure mode is fragmentation. The solution is coordination.
Emerging behaviours: The system reorganising itself
Under pressure, we are beginning to see new behaviours emerge. Hyperscalers are acting more like manufacturers, securing inputs early and standardising outputs. Governments are being drawn into system-level decisions about energy infrastructure, and contractors are being pushed towards earlier engagement and deeper integration. The traditional boundaries between client, contractor, supplier and regulator are starting to blur. This is not a planned transformation. It is an adaptive response to system stress. But it points towards a different model of construction.
From projects to systems
As we know, construction has historically been organised around projects, delivering individual assets as efficiently as possible. That model is no longer sufficient. Data centres demonstrate that success depends on factors that extend far beyond the project boundary: supply chain capacity must be secured years in advance, energy infrastructure must be aligned with development, design must reflect manufacturing constraints, and performance must be validated at the system level. The unit of delivery is no longer the project; it is the system.
A new set of relationships
Delivering at this level requires a reconfiguration of relationships among hyperscalers, original equipment manufacturers (OEMs), governments, and architecture, engineering, and construction (AEC) firms. These actors are increasingly interdependent, but not yet well coordinated. And therein lies the opportunity.
The missing capability: system coordination
As explored in my new book, Brunel’s Bees, the construction system has evolved in a way that fragments responsibility and loses knowledge between projects. The problem is not so much a lack of capability, but a lack of coordination. Data centres make this especially visible. They require continuity of decision-making, alignment across supply chains and integration of multiple systems. They expose the limits of a fragmented model.
An opportunity for transformation
This is not just a data centre issue. From 2018 to 2022, I led the UK’s programme to transform construction. Notwithstanding the impressive results we achieved, that transformation is still very much a work in progress. By learning from data centres, we can take it to the next level. If the construction sector can develop the capability to deliver these system-dependent assets effectively, that capability can be applied more broadly. Housing delivery could become more coordinated and predictable, infrastructure could align more closely with supply chains and energy and construction could become properly integrated. In this sense, data centres are a test case for the future of construction.
From delivery to system design
The implication is clear. Construction must evolve from delivering projects to designing and operating systems. That means aligning pipelines with supply chain capacity, integrating energy into development decisions, enabling programme-level coordination and supporting long-term relationships across the system. This is where a system design and integration organisation, such as A7C, of which I am a Director, is beginning to operate – helping clients and governments design the conditions under which delivery can succeed.
Conclusion: Building the hive
The AI revolution is not just a technology story. It is an infrastructure, energy, skills and governance story, and above all, a coordination story. Data centres sit at the centre of that system. They are exposing not just what we build, but how we build, and how we must change. As Brunel’s Bees argues, the future of the built environment will belong not to individual excellence, but to collective intelligence: systems that can coordinate, learn and act together. Data centres are forcing us to build that “hive”. The question is whether the construction sector is ready to lead it.
Brunel’s Bees is available from Amazon, £14.99 paperback, £9.99 eBook.
The post What data centres can teach us about transforming construction appeared first on Planning, Building & Construction Today.
