Advances in artificial intelligence (AI) have shifted data centres from a niche segment right to the fore, in terms of both demand from end users and demand from investors. So how should investors position themselves in this rapidly evolving but vitally important sector?

Of course, AI has been around for a technological “age” already. We have been guided by digital maps on our phones and anticipatory spelling in our messaging for many moons.

What is new is the evolution of the industry away from AI training of large language models, which can take place in large rural data centres, to AI inference, the application of AI. This creates a key distinction in the type of data centre because their location and their structure depends on their intended usage.

Learning and latency

The machine learning facilities that train AI “can really be almost anywhere,” explains Brian Chinappi, head of real estate and data centre investment at Actis.