Artificial intelligence is considered to be one of the most important transformational forces of our time. And yet many companies are finding it difficult to use the technology productively. The new IBM CEO Study 2025 reveals a key discrepancy: while investments in AI are increasing massively, the measurable benefits are not materialising in many cases.

For this study, the IBM Institute for Business Value, together with Oxford Economics, surveyed 2,000 CEOs from 33 countries and 24 industries worldwide. The aim was to find out how managers assess the impact of generative AI on their business models and where they are still falling behind. The study identifies five so-called ‘mindshifts’: necessary changes in thinking and behaviour that determine whether companies use AI as a genuine growth driver or run into technological dead ends.

1. Investing without a Goal

According to the study, 60 per cent of companies are still only at the pilot stage – even though many had already announced their intention to introduce artificial intelligence on a large scale in 2024 (p. 35). At the same time, 64 per cent of CEOs stated that they buy technologies before they fully understand their business benefits. These hasty investments are primarily driven by the fear of being left behind (p. 18).

This discrepancy between action and orientation is dangerous. It leads to inefficient use of resources, organisational blindness and missed opportunities. A look at the results confirms this: only a quarter of AI initiatives deliver the expected return on investment, and only 16 per cent have been scaled company-wide to date (p. 50).

At coeo, we are consciously pursuing this change with a clear technological focus. Sebastian Ludwig, CEO of the coeo Group, emphasises:

‘This is a disruptive technology that will change markets – and it affects our industry very specifically.’

2. ROI or Regression

There is great optimism: 85 per cent of CEOs expect a positive return on AI investments to increase efficiency and reduce costs by 2027, 77 per cent also for growth-oriented, scaled AI projects (p. 53). However, this belief in the technology risks being misleading if there is a lack of strategic clarity.

In fact, only 52 per cent of CEOs say that their company generates added value from generative AI that goes beyond cost reduction (p. 51). Many companies use AI to optimise existing processes – not to tap into new business potential. The good news is that a change is taking place. 65 per cent of CEOs now prioritise AI applications that clearly contribute to business results. 68 per cent have a structured system for evaluating innovation initiatives (p. 50).

The coeo Group provides an example of strategically well thought-out implementation. Kevin Yam, Chief AI Officer, explains:

‘We have built a strong team of specialists covering all topics in the field of deep learning and AI. Our goal was to develop an AI ecosystem that is specifically tailored to receivables management.’

This shows that pure lighthouse projects and proof-of-concepts are no longer enough. Companies need to make targeted investments in applications that make a clear contribution to value creation – for example through higher closing rates, optimised pricing or more precise risk analyses. The decisive factor here is not the level of innovation, but the business relevance of the solution.

3. No Breakthrough without Data

If you want to use artificial intelligence successfully, you have to start with the basics. And that means high-quality, structured and accessible data. 72 per cent of the CEOs surveyed say that proprietary company data is the most important success factor for generative AI (p. 40). In practice, however, the picture is different: half of the companies are struggling with fragmented and disconnected technology systems that make it difficult to utilise data consistently (p. 40). In addition, the reality is characterised by isolated data pools and unclear responsibilities. The study describes this situation as ‘digital debt’ – a growing backlog in the technological infrastructure that slows down change instead of accelerating it.

If you want to use artificial intelligence sensibly, you first need to lay the foundations for it: clear responsibilities, centralised data platforms, standardised interfaces and reliable governance structures.

Tamara Vrooman, CEO of the Vancouver Airport Authority, describes the transformative impact of a good data strategy as follows: ‘Data […] not only helps to eliminate challenges and points of contention, but also to identify opportunities for improvement – for new services or business areas, for example.’

The use of AI is also clearly associated with data at coeo. Sebastian Ludwig emphasises:

‘The use of artificial intelligence is predestined, especially because debt collection is a very data-driven business and we are strongly focussed on customer communication.’

4. Minds for AI

Technology alone will not bring about change if there is a lack of people who understand, apply and develop it. The study assumes that 31 per cent of the current workforce will need to be retrained within the next three years in order to meet the requirements of AI (p. 63). At the same time, 54 per cent of CEOs report that they are already recruiting for roles that did not exist a year ago.

Many companies are therefore focussing on a combination of internal training and new working models. IBM refers to this as the ‘build, buy, bot, borrow’ approach: developing talent within the company, recruiting new specialists externally, automating tasks using AI and integrating specialised partners to close skills gaps (p. 62).

Aiqing Yang, CEO of solar company JA Solar, puts it in one image: ‘As the demands of the times shift, we also need to reform our talent structure … To deal with uncertainty, we need to cultivate a circle of allies. In stormy waters, a single boat capsizes easily – but a fleet can sail far.’ (S. 65).

This point has also been understood at coeo, as Sebastian Ludwig reports: ‘At coeo, we have established a specialised AI operations unit that is being continuously expanded – with teams for performance monitoring, user experience and customer service that are specifically geared towards the productive use of AI.’

5. AI as an Accelerant

Perhaps the most significant finding of the study: 68 per cent of CEOs confirm that AI is already changing key elements of their business (p. 26). They are not just talking about more efficient processes, but about a profound change in the very essence of their company. The IBM team refers to this process as ‘AI-fuelled creative destruction’. This refers to the creative destruction of existing structures through the use of AI.

What is still considered a functioning process today may become an obstacle to innovation tomorrow. However, in order to fully utilise the potential of AI, companies need to initiate a radical rethink. Those who merely digitalise existing processes are missing the real opportunity: to rethink the business model. This also means that companies must actively switch off outdated systems in order to create space for real innovation.

Kevin Yam, Chief AI Officer, explains how coeo is helping to shape this change:

‘With cAI, we have built an integrated conversational AI platform that – based on market data and internal benchmarks – we consider to be one of the most powerful solutions in the world. What began as a solution for receivables management is increasingly developing into a generalist platform for automated communication and decision-making processes – with relevance far beyond our industry.’

It takes more than technology

The IBM CEO Study 2025 clearly shows that if you want to utilise artificial intelligence strategically, you need to realign your company. It is not the technology alone that determines success, but its embedding in clear goals, clean data structures, realistic fields of application and a motivated, empowered team.

The five mindshifts described in the study are not abstract management ideas. They are concrete calls to action. They call on CEOs to question old routines, critically review existing processes and really embrace the disruptive possibilities of AI.

Or as the study itself puts it: ‘CEOs who continue to rely on existing business, operational and governance models are stuck in the past. In today’s highly disruptive environment, this could be the biggest risk of all.’ (S. 53).