The AI Hype in Organisations: High Investment, Low Return

Organisations are allocating increasing resources to Artificial Intelligence (AI) yet persistently failing at what is truly essential to the business: capturing the return on that investment. Often driven by the fear of “falling behind” — the so-called AI FOMO (“fear of missing out”) — they invest without clearly defined objectives and without a clear ROI framework.

In a report released last month, MIT reveals that despite investments in the range of USD 30 to 40 billion in GenAI, 95% of organisations are not achieving any return. Pilot projects are proliferating, yet only 5% reach production with a real impact on the P&L (The GenAI Divide: State of AI in Business 2025, MIT NANDA). In other words, many resources allocated to innovation end up as sunk costs because they were not aligned with strategic objectives or with the organisations’ core operational challenges. The problem does not lie in the technology — which is genuinely transformative — but in the absence of a strategic approach.

To overcome this paradox, it is necessary to adopt a systematic methodology that repositions AI and advanced analytics projects around the value they effectively generate for the business. This means abandoning the logic of dispersed experimentation, where multiple pilots compete for attention and resources without clear criteria, and instead prioritising initiatives with the greatest economic impact and the lowest execution risk. The focus should be on initiatives that can be integrated into the organisation’s core processes, with clear performance metrics defined from the outset. Value capture will only then be possible through an impact assessment that accompanies the project lifecycle and reflects the organisation’s strategic priorities, including, in particular: (i) economic return (e.g., revenue growth, cost reduction, jobs created, capital invested, etc.); (ii) operational efficiency (e.g., process automation); (iii) risk mitigation (e.g., compliance, anomaly detection, etc.); and (iv) customer experience (e.g., churn reduction, personalisation, targeting, etc.). More than evaluating the technical performance of models, the objective is to ensure that each initiative contributes directly to the organisation’s strategy.

This methodological approach promotes alignment between technology teams and business areas — one of the greatest challenges of digital transformation in companies — and enables AI to be treated not as a cost or a passing trend, but as a strategic lever for growth. Moreover, monitoring the ROI of AI projects allows organisations to scale with confidence, ensuring that innovation ceases to be an experimental exercise and a cost centre, and instead generates tangible business impact.

Conversely, tracking impact from the earliest stages also enables organisations to quickly identify signs of underperformance, preventing initiatives with no results from continuing to consume resources. In short, this monitoring makes it possible to adjust priorities, redirect efforts and ensure that only projects with genuine strategic potential progress to large-scale implementation.

The greatest risk of AI is not technical failure. It is strategic failure. Isolated pilots, proofs of concept without continuity and the absence of metrics erode organisational trust, undermine team motivation and turn innovation into waste. This is not about curbing investment in AI — far from it. It is about investing strategically in AI and transforming it into a disciplined endeavour with clear metrics. Because amid the AI hype, the organisations that will truly stand out have already understood that leadership is not measured by the number of pilots or the type of technology deployed, but by direct impact on the P&L.

Original portuguese article published in Jornal Económico