AI Spend Waste: 1 in 4 Dollars Lost to Poor Governance

What happened

A new report from software delivery platform Harness has found that a significant share of enterprise AI spending is being wasted, with roughly one in four dollars allocated to AI initiatives failing to generate meaningful return. The findings, covered by CIO Dive, point to a structural governance problem rather than a technology failure: more than half of the businesses surveyed do not have a dedicated owner responsible for managing AI costs.

Without clear accountability, AI expenditure tends to sprawl — teams spin up models, tools and compute resources without central visibility, and overspend accumulates quietly. The Harness report frames this as an emerging operational risk as AI budgets grow larger and more complex across the enterprise.

Why it matters

For CX and service-design leaders, this is not an abstract infrastructure story. AI is increasingly the engine behind personalisation engines, conversational agents, sentiment analysis and real-time service routing — all of which directly shape how customers experience a brand. When AI spend lacks a named owner, the tools built on top of it are equally ungoverned: pilots stall, integrations break, and the customer-facing promise of intelligent service quietly degrades.

Behavioural economics offers a useful lens here. Diffusion of responsibility — the well-documented tendency for accountability to dissolve when it is shared across too many parties — is precisely what the Harness data describes at an organisational scale. When no single person owns the cost line, no single person owns the outcome either. That is a service-design failure before it is a finance failure.

By the numbers

• 1 in 4 dollars spent on AI is estimated to be wasted, according to the Harness report.
• More than 50% of businesses surveyed lack a dedicated owner for AI costs.

The Renascence take

Most commentary on this report will focus on procurement hygiene and FinOps tooling. That misses the deeper issue: ungoverned AI spend is a symptom of organisations that have adopted AI as a technology project rather than a customer-value project. The two framings produce very different governance structures.

When AI is owned by IT or finance, the success metric is cost efficiency. When it is owned by a CX or value-creation mandate, the metric is whether a customer's life got easier. Renascence's view is that the waste identified by Harness is not primarily a budgeting problem — it is a purpose problem. Operators who want to fix it should not start by appointing an AI cost controller; they should start by asking which specific customer moments each AI investment is designed to improve, and then assign ownership of both the outcome and the spend to the same person.