The ROI puzzle of AI investments in 2025
Artificial intelligence is set to leave a significant mark on 2025. While its potential is widely acknowledged, the road to adoption is lined with challenges. Data quality, technical debt, and uncertain returns on investment are hurdles many organizations are grappling with.
Turning AI ambitions into measurable outcomes will require not just bold investment but also a grounded approach to managing these complexities.
Insights from KPMG’s AI Quarterly Pulse Survey reveal that 67% of business leaders expect AI to fundamentally reshape their organizations within two years. Meanwhile, Forrester’s Technology & Security Predictions warn that impatience with AI ROI could lead to premature cutbacks, potentially hindering long-term benefits.
Organizations are doubling down on generative AI, with KPMG reporting that nearly 70% of leaders plan to spend between $50 million and $250 million on related initiatives over the next year. This surge in investment signals confidence in AI’s transformative potential but comes with high expectations. Despite this confidence, success metrics remain elusive. Only 31% of leaders anticipate being able to evaluate ROI within six months, and none report achieving it yet.
Traditional methods for assessing returns struggle to capture the multifaceted benefits of AI, which often include improved efficiency, customer engagement, and long-term innovation capacity. Reflecting this shift, productivity has overtaken profitability as the primary ROI metric for AI initiatives in 2025.
AI’s effectiveness depends heavily on the quality and availability of data, but this remains a persistent issue. KPMG’s survey reveals that 85% of leaders cite data quality as their most significant challenge in AI strategies for 2025.
Inadequate data not only compromises model performance but also erodes trust within organizations. Adding to this complexity is the growing technical debt caused by increasingly intricate IT landscapes, as highlighted by Forrester.
Addressing these challenges demands robust investments in data governance, infrastructure, and regulatory compliance. With the EU AI Act poised to fine companies for breaches, adherence to stringent data standards is no longer optional.
Agentic AI, which includes tools capable of performing tasks independently and adapting to real-time changes, is gaining traction. According to KPMG, over half of organizations are exploring the use of AI agents, with planned applications in administrative tasks, customer service, and content creation. Yet, only 12% have moved beyond experimentation to full deployment.
The slow adoption highlights a readiness gap that cannot be ignored. Employee engagement remains a key barrier, as only 24% of workers regularly use AI tools embedded into workflows. Seamless integration into existing systems is critical to overcoming this hurdle and achieving widespread adoption.
The rapid pace of AI development is adding to an already pressing issue: technical debt. Forrester predicts that 75% of technology leaders will face moderate to severe technical debt by 2026, as rushed implementations and fragmented systems create operational inefficiencies. This accumulation threatens to stifle innovation and complicate scalability.
AI for IT operations (AIOps) is emerging as a potential solution. These platforms can automate incident resolution, optimize system performance, and bolster resilience, offering a way to mitigate the risks associated with technical debt.
However, success with AIOps requires more than just technology—it demands a cultural shift toward greater collaboration between IT and business teams.
The journey toward meaningful AI adoption is far from straightforward, but the opportunities are too significant to ignore. Insights from recent research highlight the importance of aligning AI investments with broader business objectives while addressing critical challenges like data quality, employee adoption, and technical debt.
Leaders must address specific hurdles: