Most Enterprise AI Spend Still Haven’t Left the Lab
July 28, 2026 – 12:43 pm
Image by: KloudStax
New research reveals that most AI pilots never make it into production, with the associated spend remaining in limbo. While this might seem like failure, Jon Bitz, Chief Relationship Officer and Co-Founder at KloudStax, offers a different perspective.
The Numbers Show a Complex Picture
Forrester’s recent study on agentic AI found that three-quarters of enterprise leaders are adopting it, but only a small fraction have implemented it beyond limited pilots. Gartner predicts that by 2026, 60% of AI projects lacking data and integration infrastructure will be abandoned. Similarly, Deloitte’s survey revealed that over a third of companies utilize AI at a surface level with minimal impact on daily operations.
These findings paint a pattern: enterprises are approving AI budgets rapidly but struggling to translate that spend into functional systems. The gap between approval and implementation remains a significant concern in enterprise cloud.
The Spend is Parked, Not Wasted
Bitz believes the current situation isn’t as dire as abandonment rates suggest. He explains:
A meaningful portion of AI cloud spend is still happening in pre-production, and that’s not necessarily a bad thing. Testing, experimentation, and validation are vital steps for successful AI adoption.
However, he identifies a problem with the length of this pre-production phase, which can extend indefinitely without guidance.
Moving from Experimentation to Production
Bitz suggests focusing on converting experimental insights into production workflows rather than cutting experiment budgets. He emphasizes:
Our goal isn’t to reduce experimentation; it’s to convert it. We want to take what’s learned and turn it into production systems, shifting spend from testing into business-driving workloads.
Real Benefits: Workflow Automation
Bitz highlights that significant value is emerging in core workflow automation, where manual, time-intensive processes are transformed into efficient, production-ready systems, benefiting support operations, engineering output, and revenue-generating workflows.