Data centers and new technology loads accounted for approximately 24 percent of Ireland’s electricity requirements in 2024. That share is expected to reach 30 percent by 2032. As demand continues to rise, the relationship between data centers and the power system is changing.

Grid access can no longer be viewed simply as a question of available capacity. Increasingly, it will depend on a facility’s ability to contribute to the resilience, flexibility, and decarbonization of the network on which it depends. This is the focus of EirGrid’s proposed grid code modification: MPID345.

A new expectation for large demand

MPID345 introduces three linked requirements for large demand facilities.

1. Fault ride-through requires facilities to remain electrically connected to the transmission system during voltage disturbances.
2. Active power recovery requires facilities to restore at least 90 percent of pre-fault demand within 500 milliseconds after voltage recovers to 0.9 per unit.
3. Rate of change of frequency robustness requires facilities to withstand frequency changes of up to plus or minus one hertz per second without disconnecting.

Historically, requirements of this kind have applied mainly to generation assets, including wind and solar farms, high-voltage direct current (HVDC) interconnectors, and battery storage systems. Large electricity users faced less stringent standards because their individual effect on the system was smaller and their behavior was more varied.

That distinction is becoming harder to sustain. During a fault in January 2022, Ireland’s data centers reduced demand by 74MW, representing 16 percent of total data center demand. By May 2025, the recorded reduction had risen to 387MW, or 52 percent. When major loads respond in this way at the same time, the resulting power imbalance can worsen frequency instability across the grid.

If adopted, MPID345 would mark a fundamental change in the relationship between data centers and the power system. Large demand facilities would be expected to behave as active participants in grid stability, not simply as consumers of electricity.

The relevant boundary is the complete facility

For operators, the central point is that compliance will not be assessed asset by asset. It will be assessed by the behavior of the complete facility at the grid connection point. IT load, UPS systems, cooling equipment, variable-speed drives, generators, transformers, storage, reactive-power equipment, protection systems, and control logic all shape the active and reactive power profile seen by the network during a disturbance. What matters is how these systems perform together during the fault, transfer, and recovery sequence.

Conventional data center electrical architecture has been designed around continuity of IT load. A UPS may keep IT equipment online while its grid-side rectifier limits current, reduces input power, or disconnects in response to a disturbance. From an internal operations perspective, the facility remains online. From the grid’s perspective, however, the facility may have effectively stepped away from the system.

Operators should establish their site’s actual voltage ride-through capability, rectifier current limits, protection thresholds, battery capability, generator transfer behavior, and input-power recovery profile. Revised controls or protection settings may improve performance in some facilities. However, software changes alone cannot overcome fundamental current, thermal, or hardware constraints.

Recovery is the central challenge

Active Power Recovery is likely to be the most demanding requirement within MPID345. Restoring 90 percent of pre-fault demand within 500 milliseconds requires coordinated behavior across the whole facility.

UPS systems may recover at different rates. Mechanical systems may be restarting. Generators may be transferring. Batteries may be recharging. Protection timers may still be resetting. The demand profile seen by the grid is the combined effect of all these processes.

MPID345 is not simply an energy issue. It is a sub-second power orchestration challenge. The first question is not how much energy a facility needs. It is how much active and reactive power is required, how quickly it must be delivered, and for how long.

There is no single route to compliance. Some sites may close part of the gap through changes to UPS controls and protection settings. Others may need fast-response power support, reactive-power capability, controllable noncritical loads, or changes to the wider electrical architecture. Each approach should follow a site-level assessment supported by detailed modeling using actual network, load, protection, and OEM equipment data.

From compliance to resilience

Recovery does not end when pre-fault demand has been restored. Batteries and other energy reserves must be recharged without creating a second power step that could affect grid stability. Controlled, ramp-limited recharge is therefore an important part of a credible grid-support strategy.

Ireland may be among the first markets to formalize these expectations for large demand, but it will not be the last. As electricity systems add more renewable generation and large digital loads, dynamic performance at the grid connection will become as important as installed capacity.

The operators best placed to respond will be those that treat energy resilience as a whole-site engineering challenge. This offers a practical route to meet emerging requirements while protecting uptime, managing cost, and supporting long-term decarbonization.