From AI ambition to AI economics
Ireland’s National Digital and AI Strategy sets a clear direction for the next phase of public service transformation, writes Dennis Brown, Technology Consulting Director, PwC Ireland.
The creation of the AI Office of Ireland, the planned AI Regulatory Sandbox, and wider commitments to digital public services all point towards a more active implementation agenda. The policy signals are therefore becoming clearer. The harder question now is whether public bodies can turn growing AI adoption into measurable value, without allowing costs to drift beyond control.
This is becoming a more urgent issue as organisations move beyond experimentation. Many have already identified use cases, tested AI tools, and begun to embed them into everyday work. The discussion is shifting from capability to economics: how investments are prioritised, how benefits are measured, and how consumption is managed as usage grows.
AI changes the cost model for technology. Traditional software spending was generally built around licences, infrastructure, and implementation programmes. Costs could often be planned with reasonable confidence.
AI introduces a more variable model. Each prompt, workflow, agent interaction, and automated process consumes computing resources. At small scale, those costs may appear modest. At organisational scale, they can become material, dispersed, and harder to predict.
Recent reporting illustrates this point. a multinational car-hailing company publicly disclosed that rapid adoption of AI coding tools exhausted its annual AI budget within the first four months of the year. The deployment was not presented as a failure of the technology. The problem was the pace at which consumption-based costs rose once usage spread across the organisation.
There has also been wider concern among companies about escalating token consumption and the need for stronger controls.
Public sector organisations will not face identical circumstances. Their operating models, risk appetite, procurement routes, and service obligations differ. But the core lesson is relevant: once AI moves from pilots to routine use, leaders need visibility of both value and consumption. Without that visibility, activity can increase faster than evidence of impact.
This cannot be treated as an IT cost issue alone. Technology teams will remain central to platform selection, security, integration, and governance. Yet the drivers of AI expenditure sit across the organisation. Policy teams shape use cases. Operational teams redesign processes. Service delivery teams determine how tools are applied in practice. Individual employees influence usage every time they decide which tool to use, how much context to provide, and whether a task requires a more powerful model.
That creates a broader leadership challenge. Public bodies will need to build AI cost management as an organisational capability. This means establishing clear rules for when AI should be used, which models are appropriate for which tasks, and how benefits will be assessed. It also means ensuring that employees understand efficient use, not simply safe use.
Governance has an important role here. Much of the public debate around AI governance focuses on risk, compliance, and regulation. Those issues remain essential, particularly as Ireland responds to the EU AI Act and develops national oversight structures. The AI Office of Ireland is expected to act as the central coordinating authority for the EU AI Act, while the AI Regulatory Sandbox is intended to support supervised testing of AI solutions.
“The core lesson is relevant: once AI moves from pilots to routine use, leaders need visibility of both value and consumption.”
AI sovereignty is also becoming an important consideration. For public bodies, this means making deliberate choices about where data is held, which models and infrastructure they depend on, and how they retain sufficient control over security, resilience, governance, and cost while continuing to benefit from wider AI innovation.
However, governance should also support investment discipline. Public bodies need mechanisms to decide which AI initiatives should be scaled, which should be refined, and which should stop. In an environment where demand for AI capability may exceed available resources, those decisions cannot be made informally or left until after deployment.
Technology foundations matter for the same reason. Fragmented data, inconsistent standards, and duplicated platforms make AI harder and more expensive to scale. Better Public Services 2030 already places emphasis on interoperability, shared platforms, common data standards, and the National Data Infrastructure.
In addition to digital transformation priorities, these are also cost management priorities because reusable assets reduce duplication and support more consistent deployment.
The people dimension is equally important. PwC’s 2026 Global AI Jobs Barometer, based on more than one billion job advertisements, found that AI is reshaping work, skills, and wages, with roles enhanced by AI growing faster than those being simplified by it.
For Ireland, initiatives such as the National Skills Observatory and the Observatory for Business AI Readiness can help inform more targeted workforce planning.
The skills agenda should include cost awareness, too. Employees do not need to become technologists or procurement specialists, but they do need to understand that AI usage carries an economic dimension. Better prompts, appropriate model selection, reusable workflows, and disciplined experimentation can all influence the return generated from AI investment.
For public sector leaders, four practical priorities follow.
First, link AI investment to defined outcomes before deployment begins. Second, give each material AI initiative clear ownership and regular review. Third, build shared platforms and standards so successful solutions can be reused. And fourth, treat AI literacy as a combination of capability, judgement, and cost awareness.
Ireland has established a credible framework for AI adoption. The next phase will be judged by execution. As AI becomes more deeply embedded in public services, the organisations that manage consumption, value, and accountability together will be better placed to deliver sustainable benefits for citizens, employees, and the public finances.
W: www.pwc.ie
