As AI reshapes data centre infrastructure, the greatest threat to resilience may no longer be capacity constraints or power availability. Increasingly, it is the ability to operate and maintain complex facilities over the long term.
AI-driven demand is accelerating data centre development, increasing rack densities and placing greater pressure on power, cooling and supporting infrastructure.
The challenge, however, extends beyond building capacity. As facilities become larger and more technically complex, ensuring they remain reliable throughout their operational lifecycle is becoming equally important.
Resilience has traditionally been associated with redundant infrastructure, backup power and cooling systems. While these remain essential, they represent only part of the equation.
Maintaining resilient operations also requires established processes, effective monitoring, skilled personnel and maintenance strategies capable of adapting to changing infrastructure demands.
As the industry continues to expand, Facilities Management (FM) will play an increasingly important role in closing the gap between infrastructure capability and long-term operational performance.
Resilience is more than redundancy
A resilient data centre is not defined solely by its physical infrastructure. It depends on how effectively people, processes and technology work together to maintain consistent performance.
Redundant systems provide an essential foundation, but their effectiveness relies on well-documented procedures, experienced operational teams and the ability to identify and respond to potential risks. This is reflected in Uptime Institute’s Annual Outage Analysis 2024, which found that 80% of operators said their most recent serious outage could have been prevented through better management, processes and configuration.1
As data centres scale, operational processes must remain consistent while adapting to new technologies and changing customer requirements.
Clear documentation, knowledge sharing and reliable monitoring are therefore becoming as critical to resilience as the infrastructure itself.
Maintenance cannot wait for the calendar
Planned maintenance remains fundamental to data centre reliability, but fixed schedules alone cannot account for every operational risk.
Equipment performance changes over time, and developing issues do not necessarily align with predetermined inspection dates.
As Scott Longhurst, Data Centre Services Director at 3C 3C Datalec, explained during a recent NEDAS Live! podcast, “The maintenance schedule is not a calendar.” While maintenance activities may follow regular schedules, operators must respond to emerging performance trends and early warning signs rather than waiting for the next planned intervention.
Predictive monitoring and condition-based maintenance provide greater visibility into equipment performance, allowing teams to identify emerging issues and address potential failures before they disrupt operations.
Understanding equipment age and manufacturer recommendations is equally important. As critical systems approach later stages of their operational life, maintenance requirements may need to increase to reflect changing reliability risks.
Effective failure documentation and root cause analysis also help operators identify recurring issues and make more informed decisions about future maintenance requirements.
The objective extends beyond maintaining equipment to protecting operational continuity and managing the long-term reliability of critical infrastructure.
AI changes the maintenance equation
AI is fundamentally changing the operational requirements of modern data centres.
Higher-density computing is increasing demands on power distribution, cooling capacity and physical infrastructure. At the same time, the rapid evolution of AI hardware is shortening the timeframe in which operators must adapt to new technical requirements.
Liquid-to-chip cooling is one of the most significant developments shaping this transition.
While these technologies enable facilities to support increasingly demanding workloads, they also introduce additional operational considerations.
Unlike conventional cooling arrangements, liquid-to-chip infrastructure can involve separate primary and secondary cooling loops, with responsibilities divided between facility operators and customers.
This division introduces additional challenges around maintenance responsibilities, system monitoring and coordination between stakeholders.
Monitoring becomes increasingly important in identifying potential issues across these interconnected environments. Maintenance procedures must also evolve to reflect the specific requirements of liquid cooling infrastructure, including inspection, commissioning and ongoing system performance.
The transition to AI-ready infrastructure is therefore not simply an installation challenge. It is a long-term operational responsibility that requires specialised expertise and adaptable maintenance strategies.
Infrastructure is outpacing expertise
Data centre development is accelerating faster than the industry can train the personnel required to operate and maintain new facilities. The skills gap is becoming a resilience issue in its own right. Uptime Institute’s Global Data Center Survey 2025 found that nearly two-thirds of operators are struggling to retain staff, recruit qualified candidates, or both, highlighting the growing challenge of scaling expertise alongside infrastructure.2
The introduction of advanced cooling technologies adds another dimension to this challenge, requiring teams to develop specialist capabilities alongside established mechanical and electrical expertise.
Technical training, knowledge transfer and clear operational procedures are essential to ensuring teams can maintain increasingly complex systems safely and effectively.
Geography is also becoming a consideration. As AI infrastructure expands into more remote locations, traditional approaches to maintenance delivery may need to change.
Supporting geographically dispersed facilities requires adaptable maintenance models, workforce planning and reliable technical coverage.
The ability to scale infrastructure will increasingly depend on the industry’s ability to scale the expertise needed to maintain it.
Operational efficiency starts with visibility
Resilience and operational efficiency are closely connected.
Advances in cooling technologies, equipment efficiency and monitoring capabilities are creating new opportunities to optimise energy consumption without compromising reliability.
Modern variable-speed drives, more efficient electrical equipment and evolving cooling strategies have changed how facilities manage energy consumption.
Intelligent controls, equipment rotation and performance monitoring allow operators to optimise system operation while maintaining appropriate redundancy.
Historical performance data can help operators assess equipment reliability, determine appropriate maintenance intervals and identify when lifecycle replacements may offer operational or efficiency benefits.
Efficiency improvements must remain aligned with resilience requirements, ensuring that reduced energy consumption does not compromise operational safeguards.
The strongest FM strategies will balance efficiency improvements with the need to protect critical systems and maintain consistent service delivery.
Resilience is a lifecycle responsibility
Resilience must be considered throughout the data centre lifecycle, from installation and commissioning through to ongoing operation and maintenance.
Decisions made during the delivery phase can influence how effectively a facility operates for years afterwards.
Equipment accessibility, system integration, accurate documentation and thorough commissioning all contribute to the ability to maintain infrastructure efficiently and respond to operational issues.
For 3C 3C Datalec, this lifecycle approach reflects the company’s expanding capabilities in supporting mission-critical infrastructure.
Building on its expertise in technical installations, white space fit-outs and modular infrastructure, 3C 3C Datalec is increasingly supporting customers with ongoing Facilities Management and maintenance services.
Demand is particularly evident in AI environments, where customers require specialist support for liquid-to-chip cooling systems, including installation, commissioning, periodic inspections and ongoing maintenance.
Integrating installation expertise with ongoing operational support helps ensure infrastructure is designed, delivered and maintained with long-term reliability in mind.
As facilities become more complex, this lifecycle approach will be increasingly important in helping operators manage risk, maintain performance and adapt to future infrastructure demands.
From infrastructure resilience to operational confidence
As AI reshapes data centre infrastructure, maintaining reliable operations will become increasingly critical to long-term performance.
As power densities increase, cooling technologies evolve and facilities expand into new markets, operators must ensure that maintenance strategies, monitoring capabilities and workforce expertise develop alongside them.
The next phase of data centre resilience will not be defined by infrastructure alone. It will be defined by how effectively that infrastructure is operated, maintained and adapted throughout its lifecycle.
By Scott Longhurst
Data Centre Services Director
Explore these challenges further in Scott Longhurst’s recent NEDAS Live! podcast episode, Building Resilient Data Centers for the AI Era, where he discusses the evolving demands of AI infrastructure, proactive maintenance and operational resilience.
1 https://intelligence.uptimeinstitute.com/resource/annual-outage-analysis-2024
2 https://datacenter.uptimeinstitute.com/rs/711-RIA-145/images/2025.Annual.Survey.Report.pdf