Today’s data centre leaders face an unprecedented range of operational challenges. Traditional goals such as reducing thermal & power risk and supporting digital transformation initiatives are still critical. However, as we enter 2026, data centre operations teams are already pretty much maxxed out with the introduction of AI/HPC and liquid cooling infrastructure, while also faced with the need to unlock carbon savings as part of corporate net zero programmes and initiatives.
Addressing these activities in parallel can lead to potentially conflicting concerns. Businesses are under pressure to build AI functionality into their operations yet introducing and supporting the latest AI infrastructure can place huge pressure on data centre resilience and availability. At the same time, data centre operations are searching for ways to reduce energy usage and secure quantifiable carbon savings while simultaneously delivering against growing data centre workloads.
It’s a complex balancing act, demanding a comprehensive and sustained commitment to optimising all aspects of performance. Achieving requires new levels of insight into existing thermal performance, power provision and capacity management – levels of insight that simply cannot be achieved by relying on traditional legacy Data Centre Infrastructure Management (DCIM) and Building Management System (BMS) tools.
The visibility gap
Relying on a typical BMS view means that most data centre teams still only see their cooling unit temperatures. Rack inlet temperatures are largely unmonitored, meaning that their true status is effectively invisible. Only a small percentage of data centre M&E teams actually monitor and report equipment temperatures on a rack-by-rack basis; most data centre operations remain in the dark when it comes to effective performance optimisation.
Because of this, it’s not unusual to find expensive power and cooling resources that are being used inefficiently. This lack of real-time insight into actual data centre cooling, power and capacity performance means that operations teams often have to over-cool because of this uncertainty.
Making optimisation smarter
In our research we found that current average data centre cooling utilisation stood at just 40% – implying that significant cooling capacity was effectively stranded as nobody knew how to release it and apply it elsewhere. Even the best run data centres still have cooling, power and capacity challenges. Put an effective data centre optimisation programme into place though, and you can reduce your cooling costs by up to 30%
However, it’s not only the ability to deliver quantifiable energy and carbon savings that makes such a difference. Effective capacity planning helps you to quantify your true cooling capacity – and potentially stop unnecessary and significant spending on new cooling systems. Unlocking stranded capacity lets operations run their data centres leaner – translating directly into the kind of CO2 and cooling energy usage reductions that helps teams to reduce Power Usage Effectiveness (PUE) scores and support sustainability goals.
AI takes DC efficiency further
Taking advantage of the latest AI and machine learning capabilities enables data centre teams to take performance optimisation to the next level. Automated PUE and embedded ESG reporting can free valuable operations resources to focus on added value activity. Real-time visibility helps operations teams answer the key engineering questions that need answering before simply deploying liquid cooling – including establishing the exact blend of air and liquid cooling technologies needed within the same room.
As organisations work to make their data centre operations as efficient as possible, AI-enabled optimisation tools can also be deployed to ensure that that rooms continue to optimise cooling delivery. AI Advisory tools continuously learn about a specific cooling unit’s operation, and provide immediate advice on performance enhancements such as cooling unit changes or liquid cooling efficiency.
From reactive to proactive
Similarly many data centre teams are now taking advantage of advanced anomaly detection so that they can focus in on any drift from control set-points. They can then use the data collected from equipment such as CRACs to alert any abnormal changes in performance. Rather than wait for traditional approaches such as BMS monitoring to provide an alarm, this kind of anomaly detection can pick up on potential issues – and give the operations team time to resolve them – before they become critical.
This level of continuous innovation is essential if organisations are committed to making their data centre operations as efficient as possible. Unlocking incremental cooling, power and capacity efficiencies needs to be business-as-usual for data centre teams as they work to keep their PUE scores on a downward trajectory.
