top of page

Why is Microsoft Azure in Dublin Full?

Nov 9
13 min read
Campaigners outside the Dáil called for a moratorium on new data centres but Ireland's failure to build renewable electricity infrastructure is the real problem.
Campaigners outside the Dáil called for a moratorium on new data centres but Ireland's failure to build renewable electricity infrastructure is the real problem.

At a software engineering team meeting last week, I learned that Microsoft Azure had been unable to allocate the computing capacity we needed in Dublin for Big Red Cloud's new payroll product.


Our engineers had found a workaround so the launch wasn't in jeopardy. But they were reluctant to release computing resources when we freed them up, because they couldn't be confident of getting them back. That rather defeats the purpose of cloud computing. Cloud services are supposed to allow businesses to expand and contract their computing resources as demand changes. Payroll software is an obvious beneficiary, with its predictable bursts of activity. Instead, we were keeping capacity running as insurance against future shortages.


We are far from alone. I asked colleagues at Time Point, another Irish software business that relies heavily on Azure, whether they had experienced anything similar. They had. Their engineers had also encountered difficulties obtaining additional computing capacity and were considering reserving resources or moving some workloads to Microsoft's Netherlands region.


In July, Azure specialist Alan Kinane documented persistent capacity shortages in Microsoft's North Europe region, which is based in Dublin. Customers were struggling to obtain particular virtual machines and being directed towards other European regions. His advice included preserving existing capacity which is precisely what our engineers had been doing.


Microsoft has acknowledged pent-up demand following restrictions on data-centre expansion in Ireland. It has also announced a 17% price increase for selected Azure services in North Europe from February 2027, the largest increase among the affected regions.


The electricity infrastructure behind this growth is severely constrained. EirGrid has described the Greater Dublin network as effectively saturated for large new data-centre connections. Microsoft's own regulatory filings show that its Grange Castle expansion has had to make provision for gas-fired generation because of electricity-network constraints.

The relationship between those constraints and Azure's computing shortages is difficult to ignore. Ireland has attracted enormous demand for computing but has not developed the electricity infrastructure needed to accommodate it.


The consequences are reaching Irish businesses that depend on the cloud. I have spent years considering the other side of this equation. As a former Chair of An Taisce, I championed the organisation's policy on data centres because I believed Ireland was allowing electricity demand to grow without adequately addressing the consequences for the grid, the climate and other electricity users.


I stand over that policy. I also want Ireland to have the computing capacity needed to support businesses like mine. Those positions are not contradictory. Reconciling them requires additional renewable generation, a grid capable of carrying it and an electricity system that makes efficient use of its resources.


Our failure to deliver that infrastructure is becoming an economic and environmental problem. It also reveals something more fundamental about how Ireland makes decisions.


The wrong argument about data centres

Data centres accounted for 23% of Ireland's metered electricity consumption in 2025, according to the Central Statistics Office. A decade earlier, the figure was approximately 5%.

That is an extraordinary increase, and it has consequences. Electricity demand must be accommodated by generation, transmission and distribution infrastructure. Where renewable supply is insufficient, additional consumption can increase reliance on fossil fuels.


Concentrated demand can also impose substantial network costs. Data-centre operators should bear their fair share of those costs and meet rigorous environmental requirements.

But cloud computing is not an indulgence. Accounting, payroll, banking, healthcare, logistics and public services increasingly depend on it. Artificial intelligence will make computing infrastructure more important still.


The question is how to accommodate that demand sustainably, not whether modern economies should be allowed to have it.


An Taisce has just participated in a demonstration outside the Dáil calling for a moratorium on new data centres. The concerns raised about carbon intensity, electricity demand and pressure on the grid are legitimate. But I question where the environmental movement is directing its political energy. Data centres have become the visible symbol of a problem whose deeper cause is Ireland's failure to deliver clean electricity and the infrastructure needed to use it.


There is a further complication. The companies being targeted are also among the most important commercial backers of Ireland's renewable electricity expansion. Microsoft, Amazon, Meta and Google are major purchasers of renewable electricity through long-term power purchase agreements. These contracts give wind and solar developers greater certainty about future revenues, helping them secure financing and bring projects into operation.


Ireland's new data-centre connection policy makes this relationship more explicit. New data centres must support additional Irish renewable generation equivalent to at least 80% of their annual electricity consumption, phased in over six years.


Annual renewable procurement does not guarantee clean electricity at every hour. Nor does it solve the problem of moving electricity from where it is generated to where it is needed. Wind, solar, storage, flexible demand and transmission infrastructure must operate together.

But a blanket moratorium risks suppressing one of the strongest commercial incentives to develop additional renewable generation. Restricting demand may be necessary where the grid genuinely cannot accommodate it. It is not a substitute for expanding clean supply.


We should be marching for the delivery of renewable generation, transmission infrastructure and storage with at least as much determination as we march against the businesses that need electricity. And we should be asking why the State has been so unsuccessful in delivering them.


Permission is not delivery

John Collison has argued forcefully that Ireland has lost the ability to build. In an Irish Times essay last year, he identified fragmented decision-making, environmental restrictions and planning objections as major obstacles to infrastructure development.


His proposed remedy includes giving elected politicians greater authority over major projects, drawing on Canadian reforms intended to accelerate infrastructure approvals.

I share his frustration. I am experiencing some of the consequences.


But I think his explanation is incomplete. Some projects are delayed because responsibility is divided among too many institutions. Others are delayed because those institutions make mistakes, lack resources or fail to anticipate foreseeable requirements. Still others do not progress because politicians are unwilling to make difficult decisions, particularly when effective solutions would inconvenience influential groups. These are different failures. Concentrating authority might address the first. It will not necessarily address the others.


The ability to approve infrastructure is not the same as the ability to deliver it. Consider offshore wind. Ireland has an exceptional wind resource and ambitious plans to harness it. Offshore generation could help decarbonise electricity, reduce dependence on imported fossil fuels and accommodate growing demand from industry and data centres. Yet investors face extraordinary uncertainty about when projects will receive permission, connect to the grid and begin generating electricity.


The North Irish Sea Array, proposed off the Meath and Louth coasts, began development work in 2019. Its planning application was submitted in June 2024. In April 2025, An Bord Pleanála requested further information across 21 disciplines, including matters requiring additional marine surveys. The developer sought an extension because the work could not be completed within the original deadline. The information was eventually submitted in June 2026.


All five Phase One offshore wind projects received requests for further information. Marine environmental assessments are necessarily demanding. The effects of offshore construction on seabirds, marine mammals and habitats matter, even when the electricity produced will help tackle climate change.


But seasonal surveys cannot simply be accelerated when a deadline approaches. Their requirements should be understood and incorporated into development programmes from the outset. Were the applications sufficiently prepared? Could the planning authority have identified deficiencies earlier? Did the State provide developers with adequate clarity about what would be required?


These are questions of administrative competence and project preparation. North Irish Sea Array spent more than a year responding to a request for further information before any planning permission existed to be challenged in court. Changing judicial-review rules would not have eliminated that work.


Private equity investors have told me that capital previously available for Irish offshore wind has moved elsewhere because of the uncertainty surrounding project delivery. Equinor, Shell and Statkraft have all retreated from Irish offshore development, although their decisions reflect different circumstances and wider difficulties in the international offshore wind market.


Those wider difficulties make Ireland's performance more consequential, not less. Higher financing costs, supply-chain pressures and challenging project economics have made international investors more selective. Capital, equipment and expertise can move between countries. A developer cannot indefinitely reserve billions of euros while Ireland works out where a project may be built, how it will connect to the grid or when decisions will be made.


Time kills deals.


The limits of institutional reform

Ireland has already attempted to address fragmentation in offshore development. The Maritime Area Regulatory Authority, MARA, was established in 2023 to create a more coherent system of maritime consenting.


Its early performance was disappointing. In 2026, the Minister for the Environment criticised the authority over missed statutory deadlines and inadequate progress. Creating a specialist institution can be sensible. But the institution still needs experienced staff, effective management, clear procedures and the resources to perform its functions.


And even a better-run MARA cannot deliver an offshore wind farm on its own. Planning permission, environmental assessment, grid connections, ports, financing and construction remain the responsibilities of other organisations. The State must coordinate the whole project, not simply create another authority responsible for one stage.


Mario Draghi's report on European competitiveness makes a similar argument about the obstacles facing renewable energy and electricity-network investment. His recommendations include better-resourced permitting authorities, coordinated infrastructure planning and greater use of strategic environmental assessment.


Ireland's move towards a plan-led approach to offshore wind is promising. Identifying suitable maritime areas and their environmental constraints before individual developers commit years of work should produce better applications and more predictable decisions.


Some of Draghi's proposals for relaxing environmental requirements deserve greater scrutiny. A properly refused environmentally damaging project is a successful planning outcome, not an administrative failure. Environmental protection and public participation should be designed into infrastructure planning from the beginning. They should not be treated as obstacles to overcome after the fundamental decisions have already been made.


We should measure what actually causes delay

The North–South Interconnector offers another example of the distance between approving infrastructure and delivering it. The project will strengthen the all-island electricity system and help accommodate more renewable generation.


It has been under development for almost two decades. Its proposed overhead line has attracted sustained opposition over landscape impacts, underground alternatives and landowner rights. Independent technical reviews have reaffirmed the case for the overhead connection.


Yet practical difficulties over access to the land needed for construction have continued even after planning permission was secured. Land access is not an unforeseeable complication in building an electricity line across hundreds of properties. It is a central project requirement that should be reflected in the delivery programme from the beginning.


Greater Dublin Drainage provides a different illustration. Planning permission for the wastewater treatment project was granted in 2019 but subsequently quashed because An Bord Pleanála had failed to obtain observations required from the Environmental Protection Agency. The planning authority had made an unlawful decision. The application was returned for reconsideration. Further environmental work and consultation followed, and permission was granted again in 2025.


It is tempting to describe this as a project delayed by judicial review. But that obscures the planning authority's error, the decade or more already spent slowly developing the application, and the work subsequently required to reconsider it.


There is also a distinction between the duration of a legal challenge and the delay it actually causes to completion. In project management, the critical path is the sequence of activities that determines when a project can finish. An activity may take a year without adding a year to completion if other necessary work proceeds concurrently.


Equally, a short legal challenge can cause substantial delay if it prevents construction at a critical moment. Yet public debate routinely treats the time spent in court as though it were automatically the time added to delivery.


We need a more rigorous approach. For nationally significant infrastructure, the State should publish independently auditable critical-path histories showing changes in expected completion dates, the activities responsible and the institutions accountable. Such reporting could establish that a judicial review delayed a project substantially. It could equally reveal that land acquisition, environmental surveys, procurement, financing or administrative errors were responsible.


EirGrid already publishes transmission infrastructure delivery updates. The principle could be extended across major public projects, with independent scrutiny of the causes of delay.

Over time, this would reveal patterns. Particular agencies might consistently struggle with particular approval stages. Developers might repeatedly submit incomplete applications and appeals. Ministers might announce completion dates that were never supported by credible programmes.


It would become harder to blame others for delays that originated in one's own decisions.

The purpose would not be to eliminate disagreement. It would be to establish what actually happened, learn from it and hold the responsible institutions accountable.


Political power is not political courage

There is another problem with the argument that elected politicians simply need greater authority to approve infrastructure. It assumes that politicians will use that authority to choose effective solutions.


Galway's transport problems illustrate why that assumption deserves scrutiny. The proposed ring road has become a symbol of Ireland's inability to build. Its delays are repeatedly cited as evidence of a dysfunctional planning system.


But the city's transport problems are not simply a question of whether a particular road receives permission. Bus priority, better public transport, cycling infrastructure, traffic management and measures to manage car demand can all contribute more to reducing congestion. Some require substantial investment. Others are relatively inexpensive compared with a ring road whose estimated cost exceeds €2.1 billion.


They can also be politically difficult. Giving buses priority may mean reallocating road space. Improving cycling infrastructure may require removing parking. Managing traffic may inconvenience motorists who currently benefit from how streets are organised. These are visible, immediate decisions with identifiable opponents. A major road offers a different political proposition: a large capital project whose benefits can be promised in the future, while the difficult decisions about how people use existing roads can be postponed.


Why is the delay to a major road treated as a national failure of political leadership, while the failure to implement other potentially effective transport measures attracts so much less attention?


Politicians already possess considerable authority over road allocation, public transport priorities and investment choices. If they are unwilling to make relatively inexpensive but unpopular decisions, giving them greater authority over billion-euro projects will not necessarily produce better outcomes. The objective should be an effective transport system, not the construction of a particular road. A State that can approve megaprojects but cannot implement simpler, cheaper solutions has not solved its infrastructure problem.


The same principle applies to electricity. More transmission infrastructure is essential, but so are storage, flexible demand, fairness in cost allocation between households and large energy users, better network operation and incentives to locate electricity-intensive activity where it can be accommodated.


Participation is not the enemy

Ireland has already experimented with bypassing ordinary democratic participation in the hope of accelerating development. The Strategic Housing Development system allowed large housing applications to proceed directly to An Bord Pleanála, bypassing local authority decisions. The system produced a surge in judicial reviews and was eventually abandoned. By removing the ordinary local planning decision, the State made litigation one of the principal remaining means of challenging a permission. That experience should make us cautious about proposals to remove opportunities for public involvement.


Participation is not simply an instrument for securing public acceptance. It is a legitimate part of democratic decision-making, particularly when projects impose substantial costs on particular communities or environments. It can also improve decisions by revealing alternatives, identifying problems and challenging assumptions before developers and public authorities have committed themselves to a particular course.


One of the pillars of the Government's December 2025 Accelerating Infrastructure Action Plan is called Public Acceptance. The language is revealing. Acceptance is what a developer seeks after deciding what it wants to do. Participation can change the decision.


France's independent Commission nationale du débat public offers a more interesting approach. It oversees public debate on major projects while fundamental options can still be examined. Projects can be modified substantially, and some are abandoned. Ireland should experiment with independently governed public participation for major electricity infrastructure while fundamental choices remain open. We're doing something similar for other hot potatoes with our Citizen Assemblies.


Affected communities should have access to relevant technical and environmental expertise. Developers should have to explain how participation influenced the final decision. This will not eliminate objections or guarantee agreement. Nor should it. A meaningful process must allow a project to change or, where justified, to be abandoned. It must also allow lawful decisions to be made and implemented where legitimate interests conflict. The objective is better decisions, made earlier, with greater legitimacy.


Build the whole project

The Government's Accelerating Infrastructure Action Plan contains thirty actions addressing regulation, procurement, project delivery, coordination and public acceptance. Some are sensible. But the measure of success should not be how many actions are completed or how many new procedures are introduced. It should be whether Ireland delivers the infrastructure it needs.


Three changes would make that more likely:


First, plan the whole project. For offshore wind, that means identifying suitable development areas, environmental constraints, grid connections, port requirements and realistic construction programmes together, rather than treating them as separate problems to be resolved sequentially. For electricity-intensive development, it means coordinating additional demand with generation, storage and network capacity. Permission for a data centre is of limited value if the infrastructure needed to operate it cannot be delivered.


Second, publish the real causes of delay. Independently auditable critical-path reporting would show where projects lose time, whether reforms improve performance and which institutions need to change.


Third, make better decisions earlier. Before committing to major infrastructure, the State should demonstrate that it has considered credible alternatives, including less expensive or less environmentally damaging measures. Public participation should inform those choices while they can still be changed. And when useful measures are already within the State's powers but remain unimplemented, that failure should receive the same scrutiny as a delayed megaproject.


These reforms do not require weakening environmental protections or removing democratic rights. They require competent administration, coordinated planning and political willingness to make difficult choices.


The cost of uncertainty

Ireland needs more computing capacity. It needs renewable electricity to power that capacity, a grid capable of carrying the electricity, and arrangements that allocate costs fairly without compromising climate commitments or the interests of other users. It also needs housing, wastewater treatment, transport networks and the other infrastructure on which a functioning society depends.


The cost of failing to deliver is real, although much of it never appears in a public project's accounts. Businesses defer investment, spend money on workarounds or choose other locations. Households face constraints that should have been addressed years earlier. Renewable electricity that could replace fossil fuels remains unavailable, or cannot reach the places where it is needed. These costs are dispersed across the economy. The institutions whose decisions contribute to them rarely bear the consequences.


Our engineers have found a workaround for the moment. They are keeping Azure capacity running that they would otherwise release because they cannot be confident of obtaining it again. Offshore wind investors face a similar calculation on a much larger scale. They cannot be confident that the infrastructure needed to make their investments viable will arrive when promised. Like our engineers, they can take their capital elsewhere.


Both are rational responses to uncertainty. One wastes computing capacity. The other risks depriving Ireland of the renewable electricity it needs to decarbonise its economy. Ireland cannot work around a shortage of electricity infrastructure indefinitely.


And investors who have taken their capital elsewhere may not return when we finally decide we're ready to build.

 
 

Subscribe to get new blog posts.

bottom of page