Gartner's May 2025 survey of 506 CIOs finds that 72% report their organisations are breaking even or losing money on AI investment. The average generative AI spend was $1.9 million per organisation in 2024. The technology is not the cause of the loss. The technology performs as specified. The cause is the absence of the operating model investment that converts tool capability into organisational value. The loss is in the gap between what the technology can do and what the organisation has built to capture that capability.
The standard AI investment case covers tool costs, implementation costs and sometimes training costs. It rarely covers the cost of workflow redesign, governance design, measurement infrastructure and change management. These are the components that convert deployment into returns. Without workflow redesign, the AI tool produces individual results that plateau. Without governance design, decisions are made ad hoc and risk accumulates. Without measurement infrastructure, the organisation cannot demonstrate value to the board and the next investment cycle becomes harder to justify. Without change management, trained individuals return to untrained workflows. Each of these omissions is a gap in the investment case that turns a potentially positive return negative.
Rebuild the AI investment case to include five cost categories: tool costs, implementation costs, workflow redesign costs, governance design costs and measurement infrastructure costs. Then build the return case against operating model outcomes: workflows redesigned and in production, capability documented and transferable, value measured against pre-deployment baselines. The 72% breaking even or losing money are measuring against an incomplete investment case that never included the conditions for positive returns.
The organisations generating returns built the complete case and invested accordingly. The difference in outcome is not technology. It is the investment decision that preceded the technology.