Managing AI Spend Before It Spirals

Conal Gallagher is the CIO at both Flexera and its division Revenera, where he manages IT and information security programs.

Cloud adoption moved faster than visibility, governance and financial discipline, leaving organizations to retrofit FinOps after costs had already escalated. AI is now creating a similar challenge, with token-based consumption making spending even more volatile, distributed and difficult to forecast.

This year marked significant enterprise investment in AI, with Gartner researchers predicting that “AI spend will more than double to $5.6 trillion by 2030.” It’s no surprise, then, that traditional software budgeting models are increasingly ill-suited to the economics of AI. Unlike conventional software, every prompt, response, API call and workflow can consume tokens and generate cost. On the enterprise scale, even small usage decisions can compound quickly, making budget forecasting nearly impossible. And because AI costs can surface through cloud infrastructure, SaaS subscriptions, embedded functionality and the data required to support each use case, CIOs and other IT leaders may not have a consolidated view of where that usage, and the associated cost, is coming from.​

Instead, CIOs need to bring the same financial discipline to AI that FinOps brought to the cloud, starting with three principles that can help organizations better understand, govern and optimize AI spending.

Visibility: Know Where AI Costs Are Going

AI is being used across every level of the enterprise. But before optimization, organizations need to first understand which teams, applications, models and workflows are driving consumption.

Unfortunately, most enterprises don’t yet have a complete picture of what they’re spending on AI. Flexera 2026 data shows that only 31% of leaders believe they have accurate visibility into their AI software, and 59% stated that they had a year-over-year increase in wasted AI spend.​