Recent data centre projects are no longer being designed as scaled versions of earlier facilities. Changes in cooling systems, power requirements and internal layout point to a different set of design conditions, driven by artificial intelligence (AI) workloads.
Facilities developed by NextDC, for example, are incorporating liquid-based cooling systems to manage higher thermal loads, while hyperscale campuses led by operators including Meta are being planned at multi-gigawatt scale. In India, the entry of Adani Group into AI data centre development reflects expansion into markets where land and energy infrastructure can support large deployments.
These developments suggest a shift in how the data centre is defined as a building type.
How AI workloads are changing data centre design
The change begins with workload density. Earlier enterprise data centres operated at lower rack densities and relied on air-based cooling systems. AI infrastructure requires a different approach.
High-performance computing clusters operate at higher power densities, requiring cooling systems that move beyond air circulation. Direct-to-chip liquid cooling and immersion systems are being introduced, affecting both mechanical design and spatial organisation.
This changes the structure of the building. Floor loading increases, service zones expand and plant space takes on a larger share of the overall programme. Engineering firms such as AECOM and Jacobs are working on facilities where mechanical and electrical systems have a greater influence on the architectural response.
Why power infrastructure is shaping site selection
Power availability is now a central factor in site selection and design. AI data centres require consistent, high-capacity energy supply, often at scales that place pressure on existing grid infrastructure.
Operators including Amazon Web Services and Microsoft are aligning development strategies with access to power, leading to site selection in regions where land, transmission capacity and regulatory frameworks can support large-scale deployment.
This affects how data centres relate to their surroundings. Rather than fitting into existing urban conditions, these facilities influence regional energy planning, land use decisions and infrastructure investment.
How modular construction is accelerating data centre delivery
The pace of deployment is also affecting construction. Conventional timelines are often not sufficient for current demand.
Contractors such as Turner Construction and DPR Construction are adopting prefabricated and modular approaches. Electrical systems, cooling assemblies and rack infrastructure are manufactured off-site and installed as integrated components.
This reduces reliance on on-site labour, improves consistency and shortens delivery timelines. It also introduces a different construction approach, where parts of the building are treated as repeatable systems.
Contractors and systems shaping AI data centre delivery
Beyond developers and operators, a broader ecosystem is shaping how these facilities are designed and delivered. Engineering consultants, contractors and systems providers are increasingly central to project outcomes.
Construction firms such as Turner Construction and DPR Construction are involved in delivering hyperscale campuses, often working alongside multidisciplinary teams led by firms such as AECOM.
At the systems level, companies including Schneider Electric, Vertiv and Siemens are supplying power, cooling and monitoring infrastructure that supports high-density computing environments.
This layered ecosystem reflects a shift from single-asset development to integrated infrastructure delivery, where coordination between design, engineering and systems providers is critical.
Why data centres are becoming system-led buildings
These changes point to a shift in how data centres are assessed. Earlier facilities were often measured by floor area and capacity. Current projects are defined by compute density, energy throughput and thermal management.
Operators such as Equinix and Digital Realty are working within an environment where system performance shapes both design and operation.
This approach reflects a broader direction in building design, where operational systems determine spatial organisation.
A new typology emerging from AI infrastructure demand
As AI deployment expands, the buildings that support it continue to change. Data centres are moving into a more central role within digital infrastructure.
In this context, architecture, engineering and energy systems are closely linked. The resulting building type is defined by performance requirements, with spatial design responding to technical constraints.
