Data Centre Magazine January 2026 | Page 114

TECH & AI

60 %

Electricity costs comprise up to 60 % of operational expenses for data centre operators

30 %

Decrease in unplanned downtime through ML-powered predictive maintenance at leading facilities

40 %

Reduction in cooling energy achieved by Google’ s ML systems in data centre operations
Energy efficiency perhaps represents ML’ s most significant contribution to data centre operations. With electricity costs comprising up to 60 % of operational expenses, even marginal efficiency gains deliver substantial returns. Advanced ML systems now orchestrate cooling infrastructure dynamically, adjusting temperatures, airflow, and liquid cooling distribution based on real-time workload patterns. Back in 2016, Google’ s pioneering work demonstrated 40 % reductions in cooling energy, spurring industry-wide adoption of similar approaches.
The complexity of modern data centre networks has outpaced human capacity for manual optimisation. ML-driven network orchestration