Share with your CIO
FinOps teams, originally built to wrangle cloud costs, are fast becoming the operational backbone for AI spend governance across the enterprise. With 80% of FinOps functions already reporting to the CIO, the structural alignment is there. The problem is adoption: only 7.5% of enterprises have actually embedded FinOps into AI projects, and IDC’s Jevin Jensen estimates that gap is burning more than 15% of AI spend for 41% of companies. Meanwhile, agent deployments are forecast to grow 80x by end of 2026, from 28.8 million in 2025, turning a manageable cost problem into a sprawl crisis.
What this means for your business
The companies most exposed here aren’t the laggards who haven’t started AI projects. They’re the organizations that moved fast on AI deployment and slow on governance, which describes most of the enterprise market right now. If your FinOps team is still running cloud optimization playbooks while your business units are spinning up agents independently, you’re already behind the measurement curve. The 64% of FinOps teams now using business value delivered as a KPI, rather than cost savings, signals that the function is maturing past cost-cutting and into territory the CIO actually owns.
The shadow IT comparison Jensen raises is worth taking seriously. The reason shadow IT exploded in the cloud era was that business units could bypass IT procurement entirely, and the tools were cheap enough that no single purchase triggered a finance flag. AI agents have the same structural profile, except the downstream risks are larger because agents act, not just store. FirstRand’s Salomé Keet is threading this correctly by pulling finance and procurement data together with FinOps visibility, but that approach requires cross-functional trust that most enterprises haven’t built yet. The CIOs who get ahead of this will be the ones who treat agent inventory, a catalog tracking every deployed AI agent and its cost and purpose, as an infrastructure problem, not a policy problem.
The falsification condition for the optimistic FinOps-as-AI-governance story, one Flexera and the FinOps Foundation have clear commercial interest in telling, is whether the function can actually move fast enough. FinOps teams are small, the tooling for AI cost attribution is immature, and IDC’s 217 billion daily agent actions projection for 2029 is not a number a governance team with cloud-era headcount can manually oversee. If agent observability tooling, the systems that track what agents are doing and what it costs in real time, doesn’t mature quickly, FinOps becomes a lagging audit function rather than a live control mechanism, and the CIO’s budget exposure grows whether or not the org chart says FinOps reports to them.
Concept deep-dive: Agent inventory
An agent inventory is a centralized registry of every AI agent an enterprise has deployed, what it does, what systems it connects to, and what it costs to run. Think of it as the configuration management database that IT uses to track servers and software, applied to autonomous AI processes. Without it, finance can’t allocate costs accurately, security can’t assess exposure, and the CIO has no reliable picture of what the organization’s AI estate actually looks like. It’s the prerequisite for almost every other AI governance decision.
Based on reporting from When it comes to AI spend management, CIOs are not alone, originally published 2026-04-15 03:00:00.
