Irish firms waste EUR €720 million on failed AI projects
Large organisations in Ireland have spent an estimated €720 million on artificial intelligence projects that produced no usable outcome, according to new research commissioned by Saros Consulting.
The survey of 200 IT decision-makers found that 99% had experienced an AI project failure. Respondents reported spending an average of €770,000 on AI initiatives that delivered nothing.
It also found that 57% of organisations had spent more on failed AI projects than on successful ones. Only 58% said they had an AI strategy.
Saros Consulting based the €720 million estimate on the average reported spend on unsuccessful AI work, multiplied by the number of Irish organisations with 250 or more employees. It put that figure at 935.
Strategy gap
Pressure to adopt AI appears to be growing inside Irish companies. Nearly two-thirds of respondents (62%) said AI was important to their business.
However, the survey pointed to gaps in governance and planning. IT leaders cited regulatory concerns and budget constraints as the two most common barriers to AI implementation (24% each), while 23% pointed to a lack of AI governance.
Concerns about how AI systems make decisions also featured strongly. More than half of respondents (53%) said their organisation had implemented AI that made decisions it could not explain to customers.
Similar issues were reported with regulators: 54% said they could not explain their AI's decision-making to regulators.
The study also highlighted bias and discrimination risks. Over the past 12 months, 53% said they had discovered AI making biased or discriminatory decisions.
Leadership pressure
The findings suggest internal expectations about AI may not always match delivery realities. Some 15% said their leadership team had unrealistic expectations of what AI could achieve.
Only 58% said leadership held realistic expectations of the AI models the organisation could implement. The survey also found that 63% believed IT leaders were hiding their lack of preparedness for AI because of leadership pressure.
The results come as companies across sectors test generative AI tools and automation in customer service, compliance, software development, and knowledge work. Many are also reviewing how they manage risk, security, and accountability for AI outputs and decisions.
For Irish organisations, explainability and bias concerns also intersect with evolving regulatory and customer expectations. Firms operating across multiple markets face a patchwork of requirements on transparency, data handling, and accountability.
Consultancy view
Ray Armstrong, co-founder and co-CEO of Saros Consulting, linked the results to how organisations choose projects and set priorities.
"Our research shows that AI ambitions are coming at an unnecessary cost to enterprises in Ireland. We are seeing a pattern of businesses investing significant resources in AI without fully understanding why, and in the hope of success that often doesn't come to fruition. While the many promises of AI are undoubtedly exciting, these will remain out of reach to organisations that do not define a strategy first."
Justin van der Spuy, co-founder and co-CEO, said the pace of change in AI increased the risk of sunk costs when projects failed to translate into operational use.
"AI is ever-changing, thus demanding a steady flow of investment. While some enterprises can afford continuous expenditure on AI projects, they risk losing time and other benefits if those projects prove to be of no use. Businesses in Ireland must reassess how they decide which AI projects to invest in and ensure proper safeguards are in place. Failure to do so will result in wasted spend with no tangible return, or even worse, a negative one."
The survey was carried out by Censuswide in February 2026 among IT decision-makers in the Republic of Ireland. Saros Consulting is headquartered in Dublin and operates in the UK and South Africa.