Insight Enterprises CTO warns against unmanaged AI sprawl in enterprises

The rapid and uncoordinated adoption of AI tools within large enterprises is creating an unsustainable technical debt, according to Juan Orlandini, CTO of North America at Insight Enterprises. Orlandini warns that organizations are repeating the operational mistakes of the early DevOps era, where decentralized tool selection led to fragmented architectures and spiraling costs.

According to Orlandini, the issue stems from departmental-level purchasing and the lack of oversight on token usage. When teams act independently, they often bypass centralized cost controls, resulting in multiple, redundant AI stacks that complicate infrastructure management and balloon expense reports.

To mitigate these issues, Orlandini recommends that enterprises move toward building internal platforms, effectively creating a "golden path" for AI development. This strategy aims to limit the number of frontier models and frameworks in use, encouraging teams to build on top of established capabilities rather than starting from scratch.

Ultimately, the call is for more disciplined financial and technical governance. By applying lessons from FinOps—the practice of bringing financial accountability to cloud spending—businesses can better manage the blast radius of their AI implementations and focus resources on scalable, value-adding projects.