There is a number that should make any senior leader uncomfortable. In 2025, $547 billion of the $684 billion invested in AI globally was wasted. Not lost to fraud or mismanagement. Wasted. Invested in initiatives that produced no measurable return.
If you sponsored an AI initiative in the last two years, that number belongs to you in some proportion. So does the question that follows it: why?
The answer is available. It has been documented by MIT, Deloitte, McKinsey, and Bain across research published in 2025. Most organisations are choosing not to sit with what it says.
The failure was not technical
95% of enterprise generative AI pilots in 2025 showed zero measurable financial return. The average large enterprise lost $7.2 million per failed AI initiative.
When researchers examined why, the leading causes were not infrastructure gaps, model limitations, or integration complexity. They were:
Lack of change management
Low employee trust
No pre-defined success metrics
Loss of executive sponsorship
These are not technical failure modes. They are organisational and leadership failure modes. They are also, notably, not new. Every one of them has appeared in transformation post-mortems for the last thirty years, long before AI entered the picture.
The technology changed. The reasons for failure did not.
The framing is the variable that matters most
The second finding from the 2025 research is where the argument becomes precise.
Projects treated as business transformation succeeded 61% of the time. Projects treated as IT initiatives succeeded 18% of the time.
That is not a marginal difference. It is a 43-percentage-point gap driven entirely by how leadership chose to frame and govern the initiative from the outset. Same technology. Same sector. Profoundly different outcomes.
Comprehensive change management, applied from the beginning rather than retrofitted at go-live, drove success rates to 58%. The discipline that produced those results encompasses stakeholder alignment, adoption design, governance that connects to business outcomes, and measurement frameworks built before the first workstream launches.
None of that is new capability. It exists in organisations that have been running large-scale transformation programmes for years. The question is whether it is being deployed where AI investment decisions are being made, or whether it is sitting two floors away from the people approving the budget.
What this means for transformation leaders specifically
If you are a VP of Transformation, a programme director, or a senior leader with accountability for a major change initiative that has an AI component, this data does not describe someone else's problem.
It describes what happens when AI is handed to technology teams and treated as a deployment challenge. It describes what happens when employee readiness is addressed in week eleven of a twelve-week pilot. It describes what happens when success metrics are defined after the technology is live rather than before the business case is approved.
The organisations that succeeded treated AI adoption as exactly what it is: a change in how people work, how decisions are made, and how value is created. They applied the architecture of transformation from the beginning. Sponsor accountability, defined at the start. Adoption metrics, built into the programme structure. Change impact assessed before go-live, not after it.
That architecture is a transformation discipline, not a technology discipline. And the people who know how to build it are transformation and change professionals, not implementation partners.
The gap is organisational
The AI failure story of 2025 is not a story about technology that did not work. It is a story about organisations that applied the right technology without the governance, alignment, and adoption infrastructure needed to convert it into business value.
The gap is known. The discipline that closes it is known. The question is whether your organisation has it positioned where it needs to be, and early enough to matter.
If you are leading a transformation where this pattern is already visible, I work with senior leaders to close the gap before go-live becomes the problem.
Contact me to discuss where your programme stands.