Interconnection data centres are hubs where networks meet to exchange traffic. Telecoms operators, cloud service providers and content delivery networks use them to exchange internet traffic efficiently.

Typically located at major network crossroads, these facilities benefit from the network effects of hosting several organisations in the same location.

Hyperscale data centres are large, purpose-built facilities designed for cloud giants like AWS, Microsoft, Google and Meta. Typically single-tenant, they are engineered to meet the provider’s own technical requirements and often reach hundreds of megawatts. ​

As demand grows, hyperscale campuses are expanding into ​Tier 2 and Tier 3 geographies where land and power are more readily available.

Enterprise co-location centres allow businesses to outsource their IT workloads from on-premises (in the office) server rooms into specialist, energy-efficient third-party facilities. They support hybrid IT strategies and prioritise proximity to enterprise customers. ​

Capacity varies widely, from small sites to large 40MW facilities, depending on local demand.

AI data centres support AI inference and large language model (LLM) training.

LLM training requires large graphics processing unit (GPU) clusters and advanced cooling technology, often built in areas with abundant power. Inference workloads are more sensitive to latency and, therefore, tend to be deployed in or near cloud regions.