Kyndryl has released its third annual Readiness Report, identifying a growing gap between enterprise returns from artificial intelligence and organisations’ ability to manage the technology’s risks. While 64% of surveyed business leaders report positive returns on AI investments, average readiness for external risks has fallen to 24%, down seven percentage points from 2025. Based on responses from 3,700 senior leaders across 21 countries, the report highlights weaknesses in cybersecurity preparedness, workforce transformation, digital sovereignty and the technology foundations required to scale AI.
Kyndryl Finds Enterprise AI Returns Outpace Readiness
Enterprise artificial intelligence is moving beyond experimentation, with more organisations reporting measurable returns from their investments. But as AI becomes embedded in business operations, companies are facing a growing challenge: the systems, people and security controls needed to manage the technology are not advancing at the same pace.
That is the central finding of Kyndryl’s third annual Readiness Report. Based on responses from 3,700 senior business and technology leaders across 21 countries, the research finds that 64% of respondents report positive returns on AI investments, up 10 percentage points from 2025. At the same time, average readiness to manage external risks has fallen seven points to 24%.
The findings suggest that the enterprise AI challenge is shifting. Businesses are increasingly focused on demonstrating value from AI, but scaling those deployments safely requires more than access to advanced models or modernised infrastructure. It also demands stronger cybersecurity, workforce transformation, governance and the ability to adapt to changing competitive conditions.
AI Investment Is Delivering Returns, but Scaling Remains Difficult
The report points to a gradual transition from AI experimentation to broader business adoption. The proportion of organisations describing themselves as being in the experimentation phase fell 13 percentage points from the previous year, to 49%. Kyndryl’s report page also says 84% are deploying AI in workflows beyond pilot projects.
However, adoption does not guarantee successful implementation. Nearly half of respondents, 48%, said they had substantially changed their AI strategies during the year. Around 24% reported an AI investment that failed, while 21% said they had cancelled a major AI initiative.
These figures point to a gap between deploying AI and establishing repeatable, sustainable business value. Companies may be able to launch pilots and introduce AI into selected processes, but integrating it into mission-critical operations can expose weaknesses in data infrastructure, governance, security and organisational design.
For enterprise technology leaders, the question is increasingly how to move from individual AI projects to systems that can operate reliably across departments and business functions.
Workforce Readiness Is a Critical Part of AI ROI
One of the report’s clearest findings concerns the human side of AI adoption.
Only 27% of surveyed organisations have formally updated roles and responsibilities to account for AI. Just 29% say they have fully implemented change management for AI adoption.
The gap matters because AI can alter how work is allocated, which decisions employees make and how teams interact with automated systems. Introducing tools without revisiting responsibilities and processes can leave employees uncertain about where AI fits into their jobs and who remains accountable for outcomes.
Kyndryl found that organisations that had both updated roles and responsibilities and fully implemented AI change management were 28 percentage points more likely to describe themselves as completely ready for AI implementation. They were also eight points more likely to report positive returns on AI investments.
The result reinforces a broader enterprise technology trend: workforce transformation is not separate from AI strategy. Training, role redesign, employee engagement and operational change management can influence whether AI investments translate into measurable outcomes.
Cybersecurity Readiness Is Falling as AI Adoption Grows
Security remains the most prominent external risk in the report for the third consecutive year.
Only 26% of organisations said they were completely ready to manage cyberattacks, an 11-point decline from 2025. The finding illustrates a difficult balance: AI is helping security teams identify threats faster, but organisations are less confident about their ability to respond to attacks.
Three-quarters of respondents said AI helps their organisations identify threats more quickly. Yet better detection does not automatically translate into stronger resilience. Security teams still need incident response capabilities, reliable infrastructure, access controls and governance processes that can keep pace with increasingly sophisticated threats.
The expansion of AI agents adds another layer of complexity. Agents can interact with enterprise applications, access data and execute tasks across connected systems. Without appropriate permissions, monitoring and oversight, these capabilities can create new opportunities for misuse, data exposure or operational disruption.
For organisations scaling AI, security therefore needs to be integrated into deployment and runtime operations rather than treated as a final compliance checkpoint.
AI Is Challenging Established Business Models
The report also highlights a decline in confidence around competitive disruption from companies outside traditional industry boundaries. Readiness in this area fell 10 percentage points from 2025, to 23%.
AI agents and other emerging technologies are changing how customers discover products, interact with services and complete transactions. Retailers, for example, are considering how to benefit from AI shopping assistants without losing direct relationships with customers. Software companies face questions about the long-term economics of SaaS as AI changes the build-versus-buy equation. Financial institutions are also confronting new lending models.
These developments suggest that AI is not simply another technology investment. It can change where customer relationships are formed, how services are delivered and which companies control the interfaces through which customers access products.
Organisations therefore need to assess competitive risks beyond their immediate sectors. A company may be technically prepared to deploy AI but still be exposed if a new entrant uses AI to reshape customer expectations or undercut an established business model.
Digital Sovereignty and Infrastructure Add to the Readiness Challenge
The report also examines growing interest in digital sovereignty, particularly as governments and regulators place greater emphasis on data control and operational independence.
Kyndryl found that 68% of leaders believe solutions offering full operational control are worth paying at least a 10% premium for. Another 60% said greater control is worth accepting slower deployment and adoption of advanced capabilities.
However, interest does not always translate into action. The report identifies a gap between organisations’ stated preference for sovereignty and the steps they have taken to achieve it.
Infrastructure modernisation presents a similar contradiction. Although 71% of leaders describe their IT estates as mostly or fully modernised, 66% of CIOs believe their technology foundations require significant re-architecture before they can support AI at scale.
Modern infrastructure is not necessarily AI-ready infrastructure. AI workloads can place different demands on data access, compute capacity, network performance, security and system integration. Organisations may have upgraded individual components without addressing the architectural dependencies that determine whether AI can operate reliably across the enterprise.
Readiness Becomes an Ongoing Enterprise Capability
Kyndryl’s findings reflect a wider shift in enterprise AI strategy. The initial phase of adoption focused on experimentation, model selection and identifying use cases. The next phase is increasingly concerned with operational resilience, workforce adaptation, risk management and sustainable returns.
The report’s message is that AI readiness cannot be measured simply by the number of tools deployed or the amount invested. Companies also need to establish whether their infrastructure can support the technology, whether employees understand changing responsibilities and whether governance and security controls can manage new forms of risk.
As AI becomes more deeply integrated into business operations, the gap between adoption and preparedness could become a competitive weakness. Organisations that combine technology investment with workforce redesign, modern architecture and stronger risk controls may be better positioned to turn early AI returns into lasting business value.
Market Landscape
Enterprise AI is entering a new phase in which the focus is shifting from adoption to operational readiness. The ability to deploy generative AI, AI agents and automation tools is becoming less of a differentiator than the ability to integrate them securely and sustainably.
Four developments are shaping the market:
- AI infrastructure modernisation: Enterprises need architectures capable of supporting AI workloads, data integration, computing demands and real-time operations.
- AI cybersecurity: As AI becomes embedded in business processes, security teams must address new attack surfaces, data exposure and autonomous agent activity.
- Workforce transformation: Role redesign, employee training and change management are increasingly important to achieving measurable AI returns.
- Digital sovereignty: Organisations are weighing data control, regulatory requirements and operational independence against deployment speed and access to advanced AI services.
Kyndryl’s research captures this transition: 64% of respondents report positive AI returns, but only 24% say their organisations are completely ready to manage external risks. Its findings suggest that technology investment alone is not enough to establish enterprise AI readiness.
The competitive landscape includes cloud providers, AI infrastructure companies, cybersecurity vendors and enterprise technology services firms. Providers that can combine infrastructure modernisation, security, governance and organisational transformation are positioned to play a larger role as enterprises move AI into mission-critical operations.
Top Insights
- Kyndryl found that 64% of business leaders report positive AI investment returns, while complete readiness for external risks declined to 24%.
- Nearly half of surveyed organisations substantially changed their AI strategies, highlighting the difficulty of moving from experimentation to sustainable deployment.
- Only 27% have formally updated roles and responsibilities around AI, while 29% have fully implemented AI change management.
- Just 26% of organisations say they are completely ready for cyberattacks, despite widespread use of AI to identify threats faster.
- Modernised IT infrastructure does not guarantee AI readiness: 66% of CIOs believe significant re-architecture is still necessary.
Power Tomorrow’s Intelligence — Build It with TechEdgeAI
