Ireland’s experience with data-center expansion became less about stopping construction than about changing the terms under which large electricity users could connect. In November 2021, new connection applications faced tighter assessment criteria covering grid location, dispatchable generation or storage, and the ability to reduce demand during periods of system constraint. The resulting Dublin-area pause became a practical limit on new connections, with new Dublin-area facilities unable to secure connections until at least 2028 while applications elsewhere were assessed under the applicable connection criteria. The policy response gradually moved toward connecting investment decisions with electricity-system capability, renewable supply, flexibility, and broader economic contribution. Ireland’s later policy development shows why infrastructure approvals can consider economic activity, employment, grid efficiency, renewable additionality and wider economic benefits rather than relying on proposed megawatts alone.

From Announced Capacity to Delivered Value

A proposed large-scale facility can look economically substantial before a single server starts producing useful compute, yet its electrical capacity alone does not establish the wider economic value that the surrounding economy will receive. Ireland’s 2022 policy approach explicitly placed economic activity and employment alongside efficient grid use, renewable additionality, future-proof energy supply, decarbonization, and opportunities for smaller businesses. That matters because capacity announcements describe an asset’s intended electrical footprint, while economic value depends on what the asset enables after construction. Ireland’s experience provides a useful basis for considering those measures alongside the electrical capacity requested rather than treating proposed megawatts as the sole measure of economic contribution. The question can change from how much capacity a project promises to what measurable economic activity can reasonably accompany the capacity it receives.

The shift is particularly relevant for AI infrastructure because electrical demand can increase substantially while the permanent operating workforce at an individual facility remains comparatively limited. A site may require major transmission reinforcement, generation capacity, roads, water systems, and specialist construction services while employing a comparatively small operating team once commissioning ends. Ireland’s 2022 framework did not create a formal universal scorecard for those variables, but its policy principles clearly moved beyond electricity consumption alone by emphasizing economic activity, employment, efficient grid use, renewable additionality, and community benefits. That provides a broader basis for future approvals because proposed facilities can be assessed against grid efficiency, renewable additionality, economic activity, employment and other policy principles alongside the infrastructure they require. The same logic can apply to AI campuses elsewhere, where a megawatt allocation can be considered alongside commitments around local skills, supplier development, technical employment and productive use of compute.

The Labor-Thin Asset and the Economic Development Gap

Construction employment can make a data-center proposal appear labor intensive, even though that workforce largely exists for a defined project period rather than the operating life of the asset. Historical Irish analysis of data-center investment estimated substantial employment across construction and operations, but it separated temporary construction activity from continuing operational employment, an accounting distinction that matters when policymakers assess long-term regional development. Individual projects can show the same pattern more sharply, with one major Irish facility previously associated with about 2,000 construction jobs, while the wider Irish evidence shows that construction and operational employment represent different components of the sector’s economic footprint. That separation demonstrates why construction numbers should not substitute for an assessment of enduring employment, especially when infrastructure commitments extend beyond the construction period.

The frequently cited $1.95 million-per-job figure provides a useful warning, but it does not describe Ireland’s own data-center economics. That figure came from an analysis of 11 major U.S. data-center subsidy deals, where more than $2 billion in public support corresponded with 1,174 long-term jobs, producing an average subsidy cost of $1.95 million per job. Applying that number directly to Ireland would confuse two different policy systems, tax structures, and investment environments, so it should instead function as a stress-test metric for decision makers. Ireland could use the underlying method by calculating relevant public infrastructure and incentive costs against verified employment and wider economic contributions rather than importing the U.S. figure as an Irish benchmark. The calculation becomes more useful when it considers relevant grid expenditure, incremental tax receipts, local procurement, skills investment and measurable economic activity alongside direct employment.

Inside the Fence vs Outside the Fence

A significant share of the economic contribution associated with data-center infrastructure can occur beyond the facility boundary, where businesses and other users rely on digital infrastructure for their operations and services. A data-center operator can create a relatively small permanent workforce while supplying infrastructure that supports a much larger technology ecosystem through cloud services, advanced analytics, software development, and AI deployment. Ireland’s current policy direction recognizes this broader relationship by treating data centers as infrastructure that can support digital economic activity rather than evaluating facilities only as isolated buildings. That wider lens matters because productivity gains rarely appear on the operator’s payroll, even though they can influence investment, exports, business formation, and technical specialization across an economy. The economic case should therefore separate employment generated inside the fence from the wider economic activity enabled outside it, with each category measured independently.

That outside-the-fence value depends in part on whether businesses and workers have the capability to use digital infrastructure effectively. Ireland’s wider digital policy agenda includes measures intended to increase digital and AI capability, reinforcing the connection between digital infrastructure and the ability of businesses and workers to use it productively. This creates an important consideration for markets evaluating large compute projects because physical infrastructure alone does not guarantee that businesses and workers will capture its potential economic value. Skills spending can therefore be assessed alongside electrical infrastructure, using measures such as workforce participation, business adoption, technical capability and measurable productivity where reliable data are available. Meanwhile, infrastructure developers can strengthen their economic contribution by supporting supplier ecosystems and technical training that can remain useful beyond an individual project’s construction phase.

Accounting for the Second Subsidy

The visible incentive is only one part of the financial equation when a very large electricity load enters a constrained system, because connection, network and system requirements can also affect the overall cost of accommodating new demand. Grid reinforcement, generation adequacy, network capacity and other electricity-system requirements can create costs that do not appear in the headline value of a data-center investment. Ireland’s policy documents repeatedly identified grid capacity and security of supply as material constraints, while the 2021 connection criteria required applicants to demonstrate capabilities that could support the system during periods of constraint. That structure matters because it places additional generation, storage and renewable-energy requirements on new data-center connections instead of assuming that the wider electricity system can accommodate every new load without additional conditions. Any remaining system cost should then be identified separately rather than assumed to be part of the project’s headline economic contribution.

The scale of this issue becomes clearer when electricity demand functions as a national infrastructure requirement rather than solely as a private procurement decision. Data centers accounted for 22% of Ireland’s national metered electricity consumption in 2024, while contracted demand could raise their share to 31% by 2034 under the assumptions used in the regulatory assessment. The updated connection framework requires new facilities to provide generation or storage capacity matching their requested maximum import capacity, while additional renewable electricity requirements link future demand with new generation. Those requirements place part of the infrastructure and energy-supply obligation directly on new data-center developments rather than leaving the wider electricity system to absorb every accommodation requirement. Therefore, markets considering similar policies can disclose connection requirements, reinforcement needs, generation or storage obligations, renewable procurement commitments and demand-flexibility conditions in a common economic statement.

What Ireland Learned Before Reopening Its Market Was to Measure Return Before Booking Growth

Ireland’s experience makes more sense as a sequence of tighter connection and policy conditions than as a formal nationwide freeze followed by an unrestricted reopening. The 2021 regulatory direction did not create a formal national moratorium, although EirGrid later stopped offering connections to new Dublin-area data centers until at least 2028 because of regional grid constraints. Later policy placed greater emphasis on economic activity, efficient electricity use, renewable additionality, future-proof energy supply, decarbonization and community benefits. In 2026, policymakers moved further toward locating new large energy demand near renewable energy sources and coordinating energy-intensive investment with national infrastructure planning. That progression shows that data-center development can continue even as requirements for new grid capacity become more demanding. The underlying lesson is not that large digital infrastructure lacks economic value, but that policymakers can assess economic contribution alongside the grid and energy requirements that new capacity creates.

Roads matter because people and goods use them, power infrastructure matters because productive activity depends on it, and compute infrastructure matters because people and businesses can turn it into measurable output. Ireland’s experience suggests that other markets can apply a similar principle by examining employment, workforce capability, electricity-system costs and delivered economic outcomes alongside proposed investment. A project should not rely solely on construction employment when assessing its longer-term economic contribution, particularly when operational employment and wider economic effects can be measured separately. Ultimately, the model can treat grid capacity, energy resilience, employment, skills, productivity, tax contribution and local economic participation as connected components of one investment case. That model can give developers clearer requirements while giving infrastructure leaders a more consistent basis for assessing whether a proposed load delivers measurable economic value alongside its system requirements.