Why higher power density demands better physical infrastructure visibility
The growth of digital infrastructure is also a growth in physical infrastructure requirements. The IEA’s April 2026 assessment projects global data-centre electricity consumption to increase from approximately 485 TWh in 2025 to 950 TWh in 2030. It also identifies increasing AI-server power density and rapid changes in electrical demand. These are projections and sector-level observations, but they reinforce a practical question for individual facilities: how well can operators observe the electrical and thermal conditions supporting their computing capacity?
Distributed temperature measurements add a spatial perspective.
Power cables, cable trays and tunnels may extend beyond the locations covered by conventional equipment sensors. Published industrial applications of distributed temperature sensing describe monitoring these routes for abnormal heating and locating affected sections. This can provide an additional view of the power infrastructure, including areas that are difficult to inspect routinely.
However, a temperature measurement belongs to the location of the sensing fiber. It should not be interpreted automatically as the temperature of every conductor, connection or server nearby. A useful engineering design should therefore establish the relationship between the cable placement, the monitored asset and the alarm threshold.
The operational insight is to interpret temperature in context.
A proposed monitoring workflow should compare thermal behaviour with electrical loading, cooling conditions and operating changes. An increase associated with a planned load change should be investigated differently from a persistent localised anomaly that does not follow the expected pattern.
This approach helps frame the right question: not simply “Is the temperature higher?” but “Is the measured behaviour consistent with how this asset is being operated?”
The next opportunity is thermal decision support.
For suitable power-cable systems, real-time thermal rating combines measurements with current loading, load history and thermal models to estimate conductor temperature and permissible loading. Industry systems already use this approach. The important distinction is that additional usable capacity comes from an engineered calculation and validated operating limits—not from installing a temperature sensor alone.
Integration also needs to respect the reliability and security requirements of the facility. NIST’s guidance on operational technology explicitly covers building automation and physical monitoring systems while recognising their performance, reliability and safety constraints. Monitoring interfaces should therefore be designed as part of the operational architecture, rather than added as an uncontrolled connection to critical systems.
A practical pilot could focus on one important power route and its associated operating data. Evaluate the quality of thermal coverage, usefulness of the alarms and time required to investigate anomalies. Any improvement in uptime or maintenance cost should be demonstrated over operation, not promised in advance.
