Indian firms are earning money from artificial intelligence faster than they are learning to manage its risks. Kyndryl’s third annual Readiness Report, released in Bengaluru on 10 November, found that 60% of Indian organisations investing in AI and machine learning report positive returns, up three percentage points from 2025. Over the same period, average AI readiness across external risks fell eight points to 27%. The report is based on responses from 3,700 senior business leaders across 21 countries, including 200 in India.
The gap between returns and AI readiness is the report’s main finding. Half of the Indian organisations surveyed have changed their AI strategy in a big way in the past year, and 31% report an AI investment failure. Edelman Intelligence ran the survey between 10 July and 2 September 2026. Martin Schroeter, chairman and chief executive of Kyndryl, said AI ambitions are moving faster than the systems needed to support them. He said the bar for readiness keeps moving as AI becomes more integrated in enterprises.
AI readiness and the pilot pile-up
The clearest sign of the strain on AI readiness is the number of pilots that never reach production. Some 70% of Indian organisations say they have more AI pilots than they can realistically scale, and 58% say innovation often stalls after proof of concept. Each pilot that stops at that stage consumes budget, talent and executive attention without producing a return. It also leaves a portfolio that is expensive to maintain and hard to order by priority.
The picture is mixed. Some 85% of Indian organisations are now deploying AI in workflows beyond pilot projects, and the share still at the experimentation stage has fallen 15 percentage points to 56%. Some 80% say they are using AI agents to automate workflows that once needed coordination across several teams or systems. Moving beyond experimentation is not the same as scaling, and the report suggests many firms are caught between the two.
Why AI readiness is falling on security
Security concerns topped the list of external risks for the third year in a row and contributed most to the fall in AI readiness. Some 68% of Indian leaders identify cyberattacks as a top external concern, yet only 35% say they are completely ready to manage them, down 10 points from last year. The global figure is 26%, down 11 points. Three in four leaders say AI helps their organisations identify threats faster, even as confidence in their ability to respond has declined. Better detection is not turning into a better response.
Infrastructure is the other brake on AI readiness at scale. For the third consecutive year, Indian leaders named upgrading IT infrastructure as the top action they are taking to mitigate external risks, and 71% rate their infrastructure as mostly or fully modernised. Yet 59% say significant re-architecture is still required to support AI at scale. A system moved to the cloud and kept in good order can serve conventional workloads well and still struggle with the data access, integration and governance that AI across business processes demands.
The boardroom split behind AI readiness
Part of the problem sits inside the organisation. The report finds that 37% of Indian respondents see disagreement between the chief human resources officer and the chief technology officer or chief information officer on AI strategy. The finding matters because AI at scale is as much a workforce question as a technology one. Decisions about reskilling, role redesign and how much autonomy to give AI agents cut across both functions. When those leaders are not aligned, pilots can succeed in technical terms and still fail to change how the business operates.
The report also records a widening gap outside the firm’s own industry. Among global leaders, AI readiness for competitive disruption from outside their own industry fell 10 points to 23%, as AI agents and other technologies erode established boundaries. Software companies face questions about software as a service models as AI shifts the build-versus-buy balance.
What firms plan to do next
Schroeter’s prescription is to focus on people, technology and governance. For chief information officers facing a crowded pilot portfolio, the practical reading is to scale fewer things, more deliberately. That means choosing which pilots deserve production investment, fixing the architecture they depend on, and matching the autonomy granted to AI systems with the security to oversee them. For the 70% with more pilots than they can scale, the first test may be deciding which of them to stop. The report suggests the firms that close the gap first will be the ones that treat AI readiness as a programme of work, not a single purchase.