A new global study from HCLTech has landed a number that should stop every executive mid-presentation: 43% of major enterprise AI initiatives are expected to fail.

The “AI Impact Imperatives 2026” report surveyed 467 senior executives who are directly responsible for AI investment decisions at organisations with more than $1 billion in annual revenue. These are not early experimenters. These are companies that have committed capital, built teams, and launched programmes. Nearly half of them are on track to produce nothing meaningful.

What makes the finding genuinely important is the diagnosis. The failures are not about model quality, API access, or compute budgets. According to HCLTech, the risk is not driven by lack of experimentation or access to tools. It is driven by the difficulty of translating ambition into consistent, enterprise-wide outcomes.

The technology works. The organisations around it often do not.

The Execution Gap Is Real and Growing

Across the survey population, AI adoption is technically widespread. Enterprises are running AI initiatives in IT operations, software engineering, and core business functions. The tools are deployed. The pilots have launched. The dashboards are lit up.

But execution remains patchy. Companies move fast on tooling and slow on the conditions that make tooling useful: clear ownership, trained users, integrated workflows, and realistic timelines for impact.

HCLTech describes a shrinking window. Leaders are under intense pressure to demonstrate ROI faster than the previous wave of digital transformation. That pressure is pushing organisations to skip the foundational work and reach for the headline use case. The result is AI implementations that work in demos and fall apart in production.

What distinguishes companies that do succeed is not better technology choices. It is their ability to align ambition, execution, and accountability within tight timelines.

The People Problem Behind the AI Problem

One line from HCLTech’s research deserves to be read carefully: “The pressure to move fast is real, but without the right investment in people, in helping them understand, trust and work effectively alongside AI, speed can just as easily amplify failure as success.”

This is not a soft observation. It is describing a compounding failure mode that shows up across every large-scale technology rollout: organisations that buy capability they cannot absorb.

AI is different from previous enterprise software waves in one important way: the tools themselves require judgment to use well. An ERP system does what it is configured to do. An AI agent does what it is instructed to do, and the quality of those instructions depends entirely on how well the human giving them understands what AI can and cannot do.

Teams without data literacy make poor decisions about what to automate. Managers without AI fluency cannot evaluate whether outputs are trustworthy. Executives without practical understanding cannot set realistic expectations or make sound investment calls.

The gap is not infrastructure. It is comprehension.

What This Means for Business

If you are running AI initiatives and want to avoid being in the 43%, three questions are worth asking honestly right now.

Do your people understand the tools they are supposed to work with? Not at a surface level. At the level where they can evaluate outputs critically, identify when a model is wrong, and know when to push back rather than accept the answer.

Is someone accountable for making this work end to end? Failed AI projects almost always have sponsors but no owners. The sponsor approves the budget. The owner makes sure the thing actually runs in production and produces value.

Are your timelines realistic? The pressure to show ROI within a quarter is real. So is the fact that most meaningful enterprise AI deployments take longer to stabilise than the original business case projected. The companies that succeed plan for that and communicate it clearly upward.

The other factor the research highlights is the need for organisations to treat AI as a change management challenge, not just a technology purchase. That means deliberate upskilling, clear communication about what AI will and will not do, and feedback loops that surface problems before they become failures.

What This Means for Business Leaders

This study is a useful corrective to the prevailing narrative that AI adoption is the main challenge. For large enterprises, the adoption phase is largely done. The challenge now is execution quality.

The companies that close the execution gap share a pattern: they have invested in the foundational capability that makes AI useful rather than just present. That means people who can work effectively with AI, clear processes for deploying it responsibly, and leaders who understand what they are asking their organisations to do.

Getting that right is harder than buying another tool. It is also the only thing that separates the 57% from the 43%.

Separate research from Nasuni released around the same time reached a complementary conclusion: 97% of enterprises have deployed AI agents, but 57% of projects are failing to meet their objectives — with bad data infrastructure as the primary culprit. The two studies together paint a consistent picture: adoption is not the problem. Execution and foundations are.

---

Enterprise DNA works with organisations on both sides of the execution gap: data and AI literacy training for teams through EDNA Learn, and hands-on AI deployment through Omni by Enterprise DNA. If your AI investment is not producing the results you expected, start with a conversation.