AI Data Centers Consumed Nearly a Quarter of Ireland's Electricity in 2025

Ireland has become a testing ground for what awaits other countries betting on artificial intelligence: in 2025, the country's data centers consumed 7,663 gigawatt-hours of electricity, now accounting for 23 percent of the entire national power balance. That's a jump from just 5 percent in 2015, when data centers mainly handled cloud services and social media rather than training and running AI models.

The scale of the surge

Data from Ireland's Central Statistics Office (CSO) shows that the growth in energy appetite isn't linear, it's explosive. On a quarterly basis, data center electricity consumption rose from 291 GWh in the first quarter of 2015 to 1,991 GWh in the fourth quarter of 2025, a 584 percent increase over the decade. By comparison, urban households consume 18 percent of national electricity and rural households 9 percent - data centers have long since overtaken both segments.

Behind this growth are chiefly the biggest players in the cloud and AI markets: Microsoft, Amazon Web Services, Google, and Meta, which have placed a significant share of their European infrastructure in Ireland. The country attracted them with low corporate taxes, a cool climate favorable for server cooling, and proximity to continental markets, but that same location is now putting pressure on the national transmission grid.

A moratorium and new rules

Since November 2021, de facto restrictions have applied to connecting new data centers to the grid in the Dublin area, as the grid operator warned of the risk of supply instability. In December 2025, the regulator, the Commission for Regulation of Utilities (CRU), replaced that moratorium with a new policy for Large Energy Users. Under the policy, any facility with demand exceeding 10 MVA must have its own flexible energy source that can be switched on and off depending on grid needs, and must cover at least 80 percent of its annual energy consumption within six years from new renewable installations built in Ireland.

The measure is meant to reconcile two conflicting interests: the continued expansion of AI infrastructure, which the government and investors are counting on, with secure electricity supply for households and the rest of industry. Critics point out, however, that even under the new requirements, data centers still pay lower rates for energy than consumers - according to analyses cited in the media, households pay about 0.36 euros per kilowatt-hour, while data centers pay less than half that rate.

A microgrid near Dublin as a response

A symbol of the attempt to solve this problem is a data center launched near Dublin by Pure Data Centres Group together with AVK. It's the first facility of its kind in Europe powered by an independent, islanded energy microgrid, infrastructure that operates partly outside the main national grid. With a total capacity of about 110 megawatts and handling both cloud workloads and AI computing, the facility represents an investment worth nearly one billion euros.

The microgrid model allows the data center operator to independently manage its energy balance, easing pressure on the national transmission system during peak hours. If the approach proves successful, it could become a model for other European countries that are just entering a phase of intensive AI infrastructure expansion and are already signaling similar concerns about grid capacity.

What it means for Europe

Ireland's case illustrates a mechanism that could repeat itself in other countries betting on AI data centers: a moratorium on new connections alone doesn't solve the problem, because demand for computing power grows faster than the grid's ability to handle it. The pressure then shifts to regulators, who must require operators to build their own energy sources and contribute to the country's energy transition, rather than simply drawing on existing infrastructure.

For Poland, where a national AI gigafactory is planned and data center investment is flowing in, this is a meaningful signal - Ireland's experience shows that without early, hard requirements for operators' own power sources and renewable energy, the costs and risk of supply instability can shift within just a few years from investors onto ordinary electricity consumers.