AI Returns Rise, But Infrastructure and Talent Gaps Widen for Enterprises

Business leaders across 21 countries report improved returns on their artificial intelligence spending compared to the prior year, according to findings released Tuesday by Kyndryl. However, the same survey of 3,700 senior executives reveals a troubling gap: their ability to manage external threats has deteriorated.

The shift from pilot programs to enterprise-wide AI deployment has proven difficult for many organizations, particularly when it comes to staffing and skills. Fewer than one in three leaders indicated they have formally restructured job descriptions to account for AI-related duties or put formal processes in place to manage the organizational change that AI adoption requires.

Kyndryl CIO Kim Basile emphasized the importance of cross-functional leadership during a Tuesday media briefing. "You have to bring people along in this journey in order for them to be able to make the kind of difference with this technology that they need to," she said, stressing that CIOs cannot drive AI outcomes alone.

Security and Risk Management Lag Behind Investment

Despite rising confidence in AI's financial performance, many executives feel unprepared to manage the risks that accompany rapid technology change. Concerns center on shifting business models, aging infrastructure and mounting external threats, particularly cyberattacks.

Security emerged as the top external risk in leaders' minds, driven by high-profile incidents including OpenAI's exposure on the Hugging Face open source platform and Anthropic's Claude models leaking from their test environment.

Basile noted that while cybersecurity capabilities have advanced rapidly alongside AI, malicious actors are accelerating their own tactics. "At the same time, there are threat actors that are also moving much faster, and that technology is driving them to increase the number of threats," she said.

As systems grow more interconnected and AI assumes a larger operational role, executives need deeper insight into their infrastructure and a clearer picture of where vulnerabilities exist.

Adoption Progresses, But Foundation Work Remains

AI adoption has advanced significantly this year across enterprises. Fewer than half of surveyed leaders remain in the experimentation stage, and nearly half have substantially revised their AI strategies across technology, people and governance. One in five leaders said they shut down a major AI project after it failed to deliver.

Infrastructure readiness presents a persistent obstacle. While 71% of leaders reported their systems are nearly or fully modernized, two-thirds of CIOs believe their technical foundations require substantial upgrades before they can reliably scale AI operations.

Business leaders also expressed worry about how AI is reshaping industries beyond their own, potentially disrupting competitive dynamics and established market structures. Questions around digital sovereignty and its impact on future technology choices added to their concerns.

The Human Element Emerges as the Real Challenge

Ismail Amla, senior vice president of Kyndryl Consult, told reporters that while technical upgrades are necessary, the deeper obstacles are organizational and human. "[The challenge] is around how work changes, and whether we have got the people who are going to work with these new systems incentivized in the right way, trained in the right way and understand what it means to work with agents," Amla said.

The traditional framework of people, process and technology remains relevant, Amla explained, but AI adoption introduces layers of complexity that demand significant organizational change. Enterprises must ensure their infrastructure is secure, auditable and transparent, but the real work lies in transforming how teams operate and preparing them for new ways of working alongside AI systems.