A new report from IT infrastructure services provider Kyndryl reveals that while a majority of organizations are seeing positive returns on their artificial intelligence investments, a significant number are struggling to move their AI projects out of the experimental phase. The study, which surveyed 3,700 senior leaders across 21 countries, found that 54% of organizations are now realizing a positive return on investment from AI, a 12% increase from the previous year. However, despite this progress, 62% of these AI initiatives have not advanced beyond the pilot stage, indicating a persistent gap between initial success and widespread implementation.
The findings highlight a crucial challenge for businesses: the difficulty of scaling AI innovations. While there is a strong belief in the potential of AI, with 90% of organizations feeling they have the right tools and processes for innovation, more than half are hindered by their existing technology infrastructure. This disconnect is further compounded by a lack of workforce readiness, as less than a third of leaders believe their employees are fully prepared to leverage AI technologies. According to Kyndryl’s Chairman and CEO, Martin Schroeter, closing this readiness gap represents both the primary obstacle and the foremost opportunity for enterprises in the current landscape.
Spending Increases Amid ROI Pressure
Organizations are dedicating more financial resources to artificial intelligence, with an average spending increase of 33% since last year. This surge in investment is accompanied by heightened expectations for tangible results, as three out of five leaders report feeling more pressure to demonstrate a return on investment from AI compared to 2024. The data indicates that 68% of companies are now investing in at least one form of AI, signaling broad adoption at some level. The primary application for these investments has become cybersecurity, as businesses prioritize risk management while expanding their use of AI across various operational domains.
Confidence and Capability Disconnect
A significant discrepancy exists between leaders’ confidence in their innovation processes and the actual capabilities of their technological infrastructure. While an overwhelming 90% of leaders believe their tools and processes facilitate rapid testing and scaling of new ideas, over half of them also identify their technology stack as a major impediment to innovation. This contradiction is not new, as it mirrors the findings from Kyndryl’s 2024 report. In that study, 90% of business leaders rated their IT infrastructure as best-in-class, yet only 39% felt it was adequately prepared for future disruptions. This persistent gap underscores a fundamental challenge in aligning technological reality with strategic ambition.
Workforce Readiness and Cultural Hurdles
The report underscores a significant lag in workforce preparedness for the AI-driven transformation that a vast majority of leaders anticipate. While 87% of leaders expect AI to reshape jobs within their organizations in the next 12 months, only 29% believe their employees are equipped to use the technology effectively. This gap is not solely about technical skills; cultural factors also play a substantial role. Nearly half of the surveyed CEOs admit that their organizational culture stifles innovation (48%) and that decision-making processes are too slow (45%). These cultural and skills-related obstacles are significant contributors to the difficulty in transitioning AI projects from pilot programs to full-scale production deployments.
Pacesetter Organizations Lead the Way
The research identifies a group of high-performing organizations, termed “Pacesetters,” that approach AI readiness more holistically. These organizations place a strong emphasis on developing an adaptive culture, implementing comprehensive training programs, and ensuring strong leadership coordination, in addition to making technology investments. This integrated approach yields significant advantages. Pacesetters are 32% less likely than their peers to view their technology stack as a barrier to progress. Furthermore, they demonstrate greater resilience, being 20% less likely to have experienced a cyber-related outage in the past year, highlighting the benefits of a well-rounded strategy.
Infrastructure and Cloud Strategy Reassessment
The increasing demands of AI workloads are compelling organizations to re-evaluate their infrastructure, particularly their cloud environments. A striking 70% of CEOs describe their current cloud setup as having evolved by “accident rather than design,” suggesting a lack of strategic planning in this critical area. Geopolitical concerns are also influencing cloud strategies, with three-quarters of leaders expressing worries about the risks associated with global cloud environments. In response, 65% of organizations have modified their cloud strategies, with some opting for data repatriation, re-evaluating their vendors, or shifting towards private cloud models to ensure regulatory compliance and data sovereignty.
Scaling Challenges Persist Despite Early Success
The core challenge identified in the report is the difficulty in scaling AI projects beyond the pilot phase, even when initial results are promising. The fact that 62% of projects remain in the pilot stage, despite 54% of organizations reporting positive returns, points to systemic barriers to full-scale implementation. This stagnation occurs even as AI spending has seen a substantial 33% increase since 2024. The widespread investment in at least one form of AI by 68% of organizations indicates a broad commitment to the technology. However, the inability to move beyond the experimental stage for a majority of these initiatives suggests that financial investment alone is not sufficient to overcome the complex challenges of AI adoption.
