#United States #New York #agentic AI #AI Investments #EY Survey

Growing Investment in AI but Limited Understanding of Agentic AI

Overview

The recent wave of investment in artificial intelligence (AI) continues to show promising growth; however, organizations struggle with understanding and implementing agentic AI solutions. According to the third EY US AI Pulse Survey, while 97% of senior leaders across various industries report experiencing positive returns on investment (ROI) from their AI efforts, only a fractional percentage have effectively integrated agentic AI into their operations.

Significant Investment Trends

The EY survey, which encompasses responses from 500 decision-makers, indicates that approximately 21% of these executives have overseen investments of $10 million or more in AI, a noticeable increase from 16% just a year prior. This trend is expected to continue, with roughly a third of leaders anticipating similar or higher expenditures in the near future. Despite such optimism about the financial viability of these technologies, a considerable lag in the adoption of agentic AI technology hampers the enthusiasm surrounding its potential.

The Implementation Gap

Only 14% of senior leaders report that their organizations have fully integrated agentic AI solutions. The enthusiasm surrounding AI contradicts the tangible barriers that still exist, which include a lack of comprehensive understanding among decision-makers and the underestimation of agentic AI’s transformative capacity. While early adopters are leveraging agentic AI to optimize processes—such as enhancing customer support and improving IT efficiency—the full-scale adoption remains limited, and many organizations are left grappling with its complexity.

ROI Versus Adoption

The survey reveals that organizations investing at least 5% of their total budget in AI are reaping considerable benefits, especially in technology upgrades and enhanced customer satisfaction. A striking 86% of senior leaders say their organizations have initiated projects employing agentic AI to assist in managing processes. However, these promising results highlight a critical paradox; organizations that are well-aware of the benefits are still hesitant to fully embrace agentic AI, citing significant barriers to adoption.

Barriers to Agentic AI Progress

The survey identified several barriers to adopting agentic AI, including concerns over data security and privacy. Specifically, 35% of leaders noted cybersecurity worries, while around 30% reported apprehensions regarding data privacy. Additionally, a lack of regulations and company policies pertaining to AI usage poses challenges in integrating these technologies effectively into their existing structures. Despite these reservations, a substantial majority, 89%, expresses optimism about human oversight in operating AI systems, reinforcing the need for an integrated approach that balances automation with essential human intervention.

Future Directions

The outlook for AI remains favorable, with leaders indicating a trend towards developing in-house AI solutions rather than relying solely on acquired technologies. In fact, 64% of senior executives are focusing on building custom applications while deemphasizing the purchase of external AI-capable firms. This signifies a shift toward emphasizing tailored AI strategies that align closely with unique organizational needs, thus driving genuine innovation rather than following standard solutions.

Conclusion

While the integration of agentic AI is still catching up to the pace of investment, the findings underscore a need for organizations to bridge the gap between ambition and practical execution. Enhancing training and awareness of agentic AI's potential benefits could empower businesses to take full advantage of the promising returns on investment it offers. As they progress, embedding responsible governance and aligning capabilities with real outcomes will be crucial in navigating this new frontier.

Methodology

The EY US AI Pulse Survey involved 500 U.S. employed decision-makers spanning multiple sectors, conducted in April 2025. The survey's sample offers a margin of error of +/- 4 percentage points at the 95% confidence level. Further findings will guide leaders in their strategic approaches toward AI implementation in the future.