Condition-based maintenance (CBM) supported by artificial intelligence could significantly reduce unplanned outages and operating expenses in modern data centers. A new IDC white paper reports that early adopters of AI-backed CBM have seen up to a 75% reduction in unplanned downtime and around a 20% decrease in OpEx.
The IDC white paper is titled "The Self-Aware Datacenter: How Condition-Based Maintenance Turns Fragmented, Multi-Vendor Datacenters into Predictable Infrastructure and System Intelligence." Luis Fernandes, IDC’s lead research manager and the paper’s author, notes that higher rack densities, the integration of equipment from multiple vendors and a shortage of skilled technicians are rendering traditional, schedule-based maintenance approaches insufficient for AI-era data centers.
What has changed in data centers?
Supported by Schneider Electric, the study highlights a dramatic increase in rack power density: where traditional facilities typically averaged about 15 kW per rack, AI-focused sites can reach 300–600 kW per rack. At the same time, available electrical capacity is limited in many regions, so new compute capacity is often deployed on existing infrastructure. Operators therefore frequently inherit equipment and configurations from different vendors for which they lack operational experience, creating an urgent need to integrate those assets into performance-management systems.
The shortage of trained technical staff compounds the problem: the study cites that in the United States there are roughly seven open positions for every qualified technician. Global surveys of data centers similarly list lack of appropriately skilled personnel as a leading operational challenge.
Benefits of condition-based maintenance
According to IDC, early CBM deployments have produced tangible outcomes: fewer manual interventions, lower operating expenses and fewer unplanned outages, extended equipment lifecycles, and improved overall operational efficiency. Luis Fernandes said, "Condition-based maintenance is an operations model optimized for AI-era infrastructure. It reduces the need for manual interventions and OpEx while extending asset lifecycles."
Predictive analytics can reveal relationships between equipment from different vendors and potential failure modes, enabling operators to build machine-driven but human‑validated system intelligence.
Schneider Electric’s EcoCare and Energy Intelligence
Schneider Electric’s EcoCare services shift operations from reactive and fixed-schedule maintenance to condition-based, outcome-oriented asset lifecycle management. Combining an AI-based "Energy Intelligence" solution with expert oversight, EcoCare continuously monitors equipment and system behavior, compares it against predefined operational thresholds, detects deviations, and can predict potential failures.
Jerome Soltani, global head of Schneider Electric Services, said: "By combining remote monitoring capabilities with AI-supported coordination, we can gain a more accurate view of asset and system health and detect anomalous operating patterns early that could lead to later failures. That both helps minimize downtime durations and avoids unnecessary maintenance or shutdowns of equipment that must operate to strict specifications."
What operators can expect
The study and Schneider Electric reports identify measurable advantages from CBM adoption:
- up to 75% reduction in unplanned outages;
- approximately 20% reduction in operating expenses (OpEx) according to reports;
- reduced operational risk and extended equipment lifecycles.
The IDC white paper recommends that operators of today’s high‑density, multi‑vendor data centers consider AI‑driven condition‑based maintenance as a key approach to improving reliability, efficiency and resilience.
