CIOs in EMEA are embracing AI in an increasingly thoughtful way. This translates into more concrete objectives, fewer non-committal pilot projects, and a significantly higher transition from experiment to production.
It probably won’t surprise anyone that CIOs in Europe, the Middle East, and Africa (EMEA) still have high expectations of AI. The study into the state of Enterprise AI, conducted by IDC on behalf of Lenovo, confirms this unequivocally. In the third edition of the study, it is striking that companies are increasingly successful in distilling concrete projects, objectives, and expectations from the AI hype.
Value creation
Companies continue to see AI as a factor that creates opportunities and can drive growth. “Organizations want to continue doing what they have always done, but in a different way, in order to create extra value,” summarizes Ewa Zbrorowska, Research Director AI at IDC in Europe.
In concrete terms, this means that companies want to reinvent themselves and therefore no longer see AI as a purely IT-driven project. More and more business departments are getting a voice. AI projects are still largely situated within IT, but no longer exclusively.
Not just within IT
Within IT, IDC sees five domains in which organizations expect a lot from AI:
- IT itself;
- Cybersecurity;
- Data and analytics;
- Software development.
IDC also has a top five ready for non-IT-related divisions. It looks like this:
- CRM;
- Research and development;
- Operational tasks;
- Marketing.
Looking for (more than) return
Expectations are high. Although AI requires more and more investment, CIOs are looking forward to a return on their investments. 94 percent are optimistic that it will come. Expectations for AI within IT are the highest, followed by cybersecurity and data and analytics. CRM is the only non-IT aspect in the top five, which is concluded by software development.
IDC sees it as a sign of maturity that the return for companies is no longer measured exclusively financially. Three non-financial factors stand out: 42 percent hope for more engagement and satisfaction from employees, 41 percent are aiming for a better customer experience, and 39 percent want AI to make the decision-making process more efficient.
More adoption
That’s all well and good, but does that really make the domain of (generative) AI more mature? IDC sees adoption increasing. Where in 2024 65 percent indicated that they were still in the planning or very early phase of AI adoption, that percentage has now fallen to 43 percent. 57 percent indicated last year that they had already embraced AI, at least in the pilot phase.
However, that doesn’t tell us that much either. The sticking point in recent years has been the evolution of pilot projects, which have had a massive difficulty graduating to a fully-fledged AI solution in production.
From PoC to Production
Zbrorowska acknowledges this, and she also sees improvement there: “Last year’s research showed that barely ten percent of the proof of concept projects (PoCs) made it into production. At the time, the question was whether that was good. Given the early stage of AI adoption, ten percent was not bad for non-committal tests, but many still pointed to a loss of money.”
In 2025, almost half of the PoCs transitioned to production.
Ewa Zbrorowska, Research Director AI IDC Europe
“In 2025, almost half of the PoCs transitioned to production,” she continues. “And that is quite an improvement. It shows that organizations are successfully evolving from experimentation to use.”
There are differences on a smaller scale. For example, research from mid-2025 showed that Belgian companies find it more difficult to deploy their AI projects in production than their Dutch counterparts.
read also
From test to production and a love of hybrid: is AI coming of age?
Another trend emerges from the IDC study that Zbrorowska is happy to point out: “We see that companies are also more reasonable in setting up PoCs. The number of PoCs has decreased, which shows that organizations are thinking more carefully about how they allocate their budget and what they are testing exactly.” Zbrorowska sees this evolution as an important indication that AI is becoming more and more mature.
Hybrid AI
A second important trend emerges from the study. IDC notes that hybrid environments are preferred to support AI. Only eighteen percent of CIOs in the EMEA region want to fully embrace the public cloud. Exclusively on-premises is even less popular, with barely eleven percent indicating that they want to integrate AI in this way. Thirteen percent expect to roll out AI primarily at the edge.
That leaves the vast majority: a block of 58 percent that indicates that rolling out AI is a hybrid story, in which public, on-premises, and edge come together. Lenovo, as the sponsor of this study, will be pleased to hear that. When discussing the results, the company is only too happy to take some time to highlight its own end-to-end service provision around AI. A combination of hardware, but especially tested AI frameworks, is central to this, supported by programs that work across all sectors with room for change management.
read also
Lenovo and Nvidia Join Forces on (Hybrid) AI Agents
Simone Larsson, Head of Enterprise AI at Lenovo in EMEA, is not surprised by the preference for hybrid environments. “Organizations want to bring AI to where the data is collected,” she notes. There is still some work to be done: earlier research shows that almost half of the IT leaders in EMEA believe that the current data center infrastructure is not adequate for AI in the context of sustainability.
Management falls short
Finally, it is also about the management of AI. Organizations collectively note that good management is essential, but also that they still have some work to do in this area. More than half of those surveyed (54 percent) have to admit that they are still in the process of developing policy. Fourteen percent have developed ad hoc rules and only 30 percent indicate that they have a solid set of rules regarding AI. Two percent have nothing.
Companies are concerned about the following five risks related to AI, in descending order of importance:
- The lack of responsible AI;
- Limited data security;
- Limited knowledge about the rollout of responsible AI;
- Shadow AI and associated risks;
- Risks related to intellectual property.
IDC notes that many companies today find that they may not have all the management and policy rules ready to cover all risks as desired.
It is also striking that responsible AI and intellectual property are priorities. This contrasts with the way in which the major AI specialists are developing AI today, with an all-encompassing focus on speed.
Crucial factor
In the study, IDC notes that organizations are dealing with AI in an increasingly mature manner. The field continues to move and there is much room for improvement. A better understanding of the capabilities, clear objectives in addition to financial gain, and a preference for hybrid environments are central. Management is currently the biggest point of attention. Maturity in terms of management could become the crucial factor for the possibility of rolling out AI on a large scale, according to the researchers.