CLOUD & COLOCATION
The technology industry has long been accustomed to dealing with new waves of innovation, from the very first demands of online shopping and the emergence of software as a service, to the Internet of Things and automation. For many businesses, adapting to these rounds of digital transformations has made change a way of life.
But for data centre operators, the dual evolutions of the cloud and AI have caused particularly profound ramifications in this regard. Beyond serving as its home, cloud computing taught the data centre industry how to run itself. Orchestration, auto-scaling and self-healing systems reduced the need for constant manual oversight and became a crucial tool in the way capacity gets managed.
Marc Garner, Global President, Cloud & Service Providers at Schneider Electric, sees that legacy of automation as the foundation the industry still builds on. AI is now pushing that foundation past its original design assumptions. Cloud adoption scaled compute demand gradually, over more than a decade. AI is scaling power demand far faster, at a rate few utilities or grid planners anticipated. Data centres are becoming AI factories, and the resulting energy questions are changing how cloud providers size capacity, engage with utilities and approach facility design from the outset.
In this exclusive Q & A with Data Centre Magazine, Marc argues that the same automation thinking behind cloud infrastructure can be redirected towards AI energy demand itself. He sets out why meeting AI’ s energy demands is also an opportunity – one that could build digitalised, AI-enhanced grids in which large energy users become active partners rather than passive consumers, powering the AI revolution responsibly and sustainably.
Q. WHAT DOES THE NEW ENERGY LANDSCAPE LOOK LIKE FOR DATA CENTRES?
» According to IEA estimates, by 2030, data centres could consume 7-10 % of electricity in the US, and up to 3 % globally. In Ireland, this could be up to 30 %. Almost two-thirds( 60 %) of this new demand is predicted to be AI‐driven.
As AI develops and expands, model training is driving extreme equipment density, with power consumption profiles differing greatly from previous generations of accelerated computing workloads. Moving beyond large language models( LLM), there are expectations that vertical AI, where models are developed for specific verticals, will drive further adoption.
These rising power demands have seen Gartner predict that 40 % of existing AI data centres will be operationally constrained by power availability by 2027. However, sustainability goals remain achievable through the relentless pursuit of efficiency in equipment, systems and design, and with developments in liquid cooling and AI-powered energy optimisation.
68 August 2026